Significance: Since the turn of the 21st century, artificial intelligence (AI) has advanced considerably in many domains, including government affairs. Furthermore, the emergence of deep learning has taken the development of many AI fields, including natural language processing (NLP), to a new level. Language models (LMs) are key research directions of NLP. Referred to as statistical models, LMs were initially used to calculate the probability of a sentence; however, in recent years, there have been substantial developments in large language models (LLMs). Notably, LLM products, such as the generative pretrained transformer (GPT) series, have driven the rapid revolution of large language research. Domestic enterprises have also researched LLMs, for example, Huawei's Pangu and Baidu's enhanced language representation with informative entities (ERNIE) bot. These models have been widely used in language translation, abstract construction, named-entity recognition, text classification, and relationship extraction, among other applications, and in government affairs, finance, biomedicine, and other domains. Progress: In this study, we observe that improving the efficiency of governance has become one of the core tasks of the government in the era of big data. With the continuous accumulation of government data, traditional statistical models relying on expert experience and local features gradually suffer limitations during application. However, LLMs, which offer the advantages of high flexibility, strong representation ability, and effective results, can rapidly enhance the intelligence level of government services. First, we review the research progress on early LMs, such as statistical LMs and neural network LMs. Subsequently, we focus on the research progress on LLMs, namely the Transformers series, GPT series, and bidirectional encoder representations from transformers (BERT) series. Finally, we introduce the application of LLMs in government affairs, including government text classification, relationship extraction, public opinion risk identification, named-entity recognition, and government question answering. Moreover, we propose that research on LLMs for government affairs must focus on multimodality, correctly benefit from the trend of "model as a service, " focus on high data security, and clarify government responsibility boundaries. Additionally, a technical path for studying LLMs for government affairs has been proposed. Conclusions and Prospects: The application of LLMs in government affairs mainly focuses on small-scale models, lacking examples of application in large-scale models. Compared with smaller models, large models offer many advantages, including high efficiency, broader application scenarios, and more convenience. These advantages can be understood as follows. In terms of efficiency, large models are usually trained on a large amount of heterogeneous data, thus delivering better performance. In terms of application scenarios, large models gradually support multimodal data, resulting in more diverse application scenarios. In terms of convenience, we emphasize the "pretraining + fine-tuning" mode and the invocation method of interfaces, making LLMs more convenient for research and practical applications. This study also analyzes the issues suffered by LLMs, specifically from the technological and ethical perspectives, which have resulted in a panic to a certain extent. For example, ChatGPT has generated many controversies, including whether the generated files are novel, whether using ChatGPT will lead to plagiarism and ambiguity as to who are property rights owners for the generated files. Overall, it can be said that LLMs are in the stage of vigorous development. As the country promotes research on AI and its application in government affairs, LLMs will play an increasingly crucial role in the field.

Objective: The resilience of transportation networks is a prominent research area in transportation safety. However, current studies on transportation network resilience often inadequately measure the changes in spatiotemporal travel costs for passengers, primarily focusing on the recovery phase rather than the resistance phase in two-stage resilience. There is also insufficient identification and analysis of critical segments, and a lack of suitable resilience simulation and evaluation methods for urban agglomeration railway passenger transport networks. This paper proposes a resistance resilience assessment model and a resistance resilience simulation evaluation process for urban agglomeration railway passenger transport networks centered on spatiotemporal accessibility for passengers. The aim is to evaluate the resistance resilience of these networks and identify critical segments. Methods: This paper explores the concepts of resistance resilience and recovery resilience within transportation networks. Utilizing the complex network Space L modeling method, this paper develops a spatiotemporal weighted urban agglomeration railway passenger transport network model that considers actual railway passenger stations as network nodes. Segment interruption scenarios were simulated using attack modes involving single segment deletion and multiple segment continuous deletion. A dynamic resistance resilience evaluation index termed the network performance retention rate, was introduced based on the performance response function and spatiotemporal accessibility of passengers. This paper devises a resistance resilience assessment model and simulation evaluation process to evaluate the substitutability of segments and the overall network resistance resilience. The Chengdu—Chongqing urban agglomeration was selected as a case study to identify and compare critical segments and resistance resilience across unweighted, spatially weighted, and temporally weighted railway networks. Results: The results of this paper were as follows: (1) The interruption of critical segments near railway hub cities could lead to a maximum network performance loss of 12.23%. It was necessary to identify critical segments through predisaster simulations. (2) Significant differences were found in the critical segments identified through resistance resilience simulations across unweighted, spatially weighted, and temporally weighted railway networks. The Spearman correlation coefficient indicated a relatively poor correlation between the critical segment rankings of unweighted and weighted railway networks. (3) The resistance resilience indices of the three railway networks highlighted that single segment interruptions significantly affected travel time. (4) Continuous interruption of identified critical segments severely affected network performance, with temporally weighted railway networks experiencing a stronger impact than spatially weighted and unweighted railway networks. Predisaster simulations solely based on topological structure or spatial distance might underestimate the consequences of risk interference. Conclusions: The methods proposed in this paper address the gap in targeted research on the resistance resilience of railway passenger transport networks in urban agglomerations. Simulations of single segment interruption and multiple segment continuous interruption enable the identification and verification of key network segments. Additionally, analyzing the network resistance to interruptions provides a scientific foundation for transportation network planning and decision-making. Furthermore, analyzing the network's resilience evaluation index of the network performance retention rate proposed in this paper offsets the impact of disturbance time uncertainty, providing a scientific foundation for transportation network resilience research.

Objective: Implementing differentiated toll discounts for expressway trucks can lead to a more balanced traffic flow across the road network. Expressway operators hope to make profits by charging truck tolls, while truck groups aim to maximize profits through toll charges, whereas truck groups focus on minimizing travel costs in terms of economics and time. A balance exists between the benefits of both parties; however, determining differentiated toll discounts for expressways to reach this balance is difficult. Methods: (1) Based on consumer surplus theory, key factors affecting freight route selection are identified using the preference survey of traveling behavior, and a bi-level programming model is proposed for determining differentiated toll discounts, incorporating assumptions and constraints. (2) The upper model, considering truck cost and travel time, is a surplus maximization model for expressway operators. It is solved using a novel algorithm enhanced by combining a genetic algorithm with simulated annealing. In the upper model, a lower limit on the financial revenue targets of highway operating enterprises is included the constraints to avoid overflow of lower bound returns during the iteration process. (3) The lower model leverages a logit-based stochastic user equilibrium allocation model for multiple vehicle types under elastic demand, solved using the Frank Wolfe algorithm. A generalized impedance function considering economic and time costs is established in the lower model to demonstrate the impacts of road conditions on truck travel. Cost weighting coefficients are introduced, and calculation methods and recommended values are proposed to integrate economics and time costs. (4) Detailed execution steps are provided for solving algorithms of the upper and lower models. The model also introduces model convergence criteria to optimize the iteration efficiency of the solving algorithm. A fitness function is proposed based on the financial lower bound target, and the upper model is transformed into a minimum value problem, eliminating the constraint of discounted rates. Results: The feasibility of the model is validated using toll collection data of expressway and link traffic data of highways, with three instance highway sections. A reasonable range suitable for implementing differentiated toll discounts can attract trucks back to the expressway, and increasing the daily average traffic volume for each vehicle type. After 43 iterations, the upper model achieves a stable function value. The toll discount rates for small trucks, medium trucks, heavy trucks, and extra-heavy trucks on the instance expressway fall within the ranges of 78.68%-86.27%, 55.82%-65.82%, 47.90%-54.81% and 47.52%-48.31% respectively; consequently, the average truck flow on the expressway increases by 12.24%. Conclusions: The conclusion demonstrates that the bi-level programming model can accurately determine the toll discount range for trucks on expressways; however, even with a discount rate of 4.7% for oversized trucks on nearly 100 km of the actual expressway, attracting all oversized trucks to return to the expressway remains challenging. Fuel and toll fees remarkably impact travel path selection within the generalized impedance function; moreover, the same toll discount produces notable differences in implementation effects across truck types. The research provides support for developing differentiated toll policies for expressways, as well as their subsequent optimization and adjustment.

Objective: Spontaneous coal combustion is one of the major natural disasters in coal mining; thus, accurate prediction of the risk of spontaneous coal combustion is crucial to prevent and control coal fire disasters. However, the complexity of the physicochemical process of spontaneous coal combustion and its various influencing factors poses a challenge to the risk prediction of spontaneous coal combustion. Strengthening research on spontaneous coal combustion hazard prediction technology using deep learning is crucial for improving the intelligent control level of coal mine safety production. Methods: In this study, CO volume fraction was chosen as the index for spontaneous coal combustion evaluation. A dataset was constructed, and the field observation data were visualized. Next, the dataset was tested for the distribution of eigenvariables, normalized for the distribution of eigenvariables, and normalized for the dataset using kernel density estimation, logarithmic transformation, and maximum-minimum normalization. Finally, three algorithms, namely recurrent neural network (RNN), long short-term memory (LSTM) network, and gated recurrent unit (GRU), were applied to the data mining of spontaneous coal combustion feature information, and a dynamic sequence prediction model of spontaneous coal combustion CO volume fraction was established. During the model construction process, the full connectivity layer and Dropout class were added to prevent overfitting, and the mean square error and three model performance test indicators were introduced to analyze and optimize the model parameters and test the model performance. Results: The results were presented as follows: (1) The CO volume fraction sequence dataset was established based on the field data of the Dafosi Coal Mine, the model generalization capability was enhanced, and the training time of the model was shortened by preprocessing the dataset. (2) The RNN, LSTM, and GRU models achieved the dynamic prediction of CO with an error of less than 1 %. (3) The optimal parameters of the three models were determined from the mean absolute error (MAE), the root mean square error (RMSE), and R² of the training and validation sets. A comparative study using the model performance evaluation metrics revealed that the LSTM model had the highest prediction accuracy under the same sequence data, followed by the RNN and GRU models. Conclusions: Using 285 sets of field data, the spontaneous coal combustion CO volume fraction sequence prediction models based on the RNN, LSTM, and GRU algorithms were established. The experimental values of the CO volume fraction were highly consistent with the predicted values, and the prediction error was less than 1 %. The model can predict the change in the CO volume fraction in future moments using the dataset. The results reveal that the dynamic time series prediction of CO volume fraction from spontaneous coal combustion using sequence models is possible compared with conventional static models. Moreover, the process of constructing the three models and optimizing the parameters can be employed as a basic study for developing sequence prediction models for other indicator gases.

Objective: Formations of fixed-wing unmanned aerial vehicles (UAVs), which are commonly used in military, rescue, and other missions, often do not have the ability to hover and have a large turning radius. Thus, when operating in an unknown environment, it is easy for the formations to collide in the presence of obstacles, which will gravely affect flight safety if not guarded against. It is difficult to avoid unknown environmental obstacles using traditional modeling methods. However, artificial potential field methods can address deadlock problems such as target infeasibility and cluster congestion. Methods: To achieve the cooperation of UAV formations without collision, a deep deterministic policy gradient (DDPG)-based centralized UAV formation control method is proposed in this study, which is designed by combining the centralized communication architecture, reinforcement learning, and artificial potential field method. First, a greedy-DDPG flight control method is studied for leader UAVs, which improves collision avoidance effectiveness. Considering maneuver constraints, reward functions, action spaces, and state spaces are improved. Additionally, to shorten the training duration, the exploration strategy of DDPG is improved using the greedy scheme. This improvement mainly uses the critic network to evaluate the value of random action groups and improves greedy selection to make actions more inclined, thus achieving rapid updates regarding the critic network and accelerating the update of the overall network. Based on this, incorporated with the artificial potential field method and leader-follower consensus, a collision-free control method is designed for followers, which can ensure collision-free following cooperation. Results: The numerical simulation experimental results show that the improved DDPG algorithm has a 5.9% shorter training time than the original algorithm. In the same scenario, the method that we proposed perceives the same number of obstacles as the artificial potential field method. The artificial potential field method has significant fluctuations in heading angle, while the proposed method has relatively small fluctuations. The DDPG algorithm has a smoother heading angle due to a smaller number of perceived obstacles; however, the minimum distance from the obstacles is only 9.1 m. The method that we proposed here is above 17 m from the obstacles. Furthermore, Monte Carlo experimental data under different scenarios of the long aircraft show that the ability of obstacle avoidance generalization of the proposed method is improved. Moreover, experiments were applied to the proposed formation control method. Under the same scenario and control parameters, the UAV formation control method based on the proposed architecture has lower formation errors during flight, with a maximum error of no more than 10 m. However, the artificial potential field-based formation control method has a maximum formation error of over 25 m. When encountering narrow gaps, our proposed method can quickly pass through without congestion, while the artificial potential field-based formation control method appears to hover in front of obstacles, which is not conducive to flight safety. During the entire flight, this method has a greater distance from obstacles and higher safety. Conclusions: Compared with the original DDPG algorithm, the improved DDPG algorithm has faster training speed and better training effect. The formation control method can realize the formation flight of unmanned aerial vehicles under unknown obstacles. Compared with the formation control method based on artificial potential field, the formation control method avoids the hovering in place before obstacles, which is of great significance to the formation flight safety of unmanned aerial vehicles.

Objective: Saline-alkali soil is an important reserve resource of cultivated land and potential granary in China, and its management and utilization are related to national food security. Therefore, innovative techniques and amendments should be developed to address these challenges in saline-alkali regions. Among these, calcium supplementation is recognized as one of the most effective methods for ameliorating saline-alkali soil. In the past two decades, gypsum from the desulfurization of flue gas (FGDG) in coal-fired power plants has become a preferred calcium source for ameliorating saline-alkali soil because of its high calcium content and economic feasibility. Given that FGDG has developed into a soil amendment and has been widely used, a profound understanding of the progress of its patents can provide technical guidance for the large-scale amelioration of saline-alkali soil. Methods: Based on the incoPat global patent database, a bibliometric analysis was conducted on 520 invention patents in the field of using FGDG to ameliorate saline-alkali soil from 2003 to 2022. The application and authorization trends, high-yield mechanisms, operational status, substance composition, and their correlation with patents in this field were systematically analyzed. In addition, a comparative analysis was conducted on the effectiveness of 52 patents with application cases. Results: The results showed that the annual number of patent applications for using FGDG amendments to ameliorate saline-alkali soil has a trend of first increasing and then decreasing, with a peak period of 115 patents in 2016. Most patents take 20-30 months from publication to authorization. However, the overall proportion of authorization has shown a decreasing trend. The number of patents granted by universities and research institutes is higher than that granted by enterprises, whereas the number of patents jointly granted by universities and enterprises accounts for 15.6% of the total. A total of 37 patents were converted, 7 of which were pledged, accounting for 33.3% of the total number of grants, all of which were transferred by universities to enterprises and pledged by enterprises for financing. More than 70% of patents comprised three or more substances, primarily including organic and inorganic minerals, microbial agents, and nutrient supplements. Organic materials can directly provide nutrients for the soil to make up for the shortage of FGDG in terms of nutrients, with the frequency of application as high as 95.7%, followed by inorganic minerals, which account for 44.5%; microbial agents, which account for 41.3%; and nutrient supplements, which account for 21.3%. Compared with soils with or without other types of amendments, the application of FGDG amendments significantly decreased soil pH, exchangeable sodium percentage, and salt ions that are toxic to crop growth and increased soil Ca²⁺, SO₄^2-, and total/available nitrogen and phosphorus contents, which provided a better soil environment, thereby increasing crop yield. Conclusions: Generally, research and development on FGDG amendments for saline-alkali soil amelioration have matured, and some innovative achievements have been transformed into real productivity; thus, the value of related patents has been increasingly highlighted. However, problems such as the relatively simple composition of current patents, unclear technical requirements for the amount of application and method, and serious homogeneity of patents have been encountered. In the future, we should strengthen the cooperation among schools, enterprises, universities, and research institutes, intensify research on the FGDG formula used in saline-alkali soil, and enhance the application benefits of FGDG amendments.

Objective: The atmosphere-breathing electric propulsion (ABEP) system has become a highly promising candidate for drag compensation in spacecraft operating in very low Earth orbit. To improve the inlet design of ABEP systems, this study performs a comprehensive numerical investigation of gas flows inside the inlet. The primary objective is to gain insight into the effects of the gas-surface interaction (GSI) model on the flow features, compression, and collection performances. Methods: This paper explores ABEP inlet flows using the direct simulation Monte Carlo (DSMC) method. A typical altitude of 180 km in the upper atmosphere is considered, and four GSI accommodation coefficients (σ=1, 0.8, 0.5, and 0.2) are selected. The DSMC method simulates gas flows according to the motion of a cluster of simulation particles, where each particle represents a large number of real gas molecules. In the DSMC method, particle motions are computed deterministically, whereas intermolecular collisions are calculated statistically. Each simulation particle travels at a constant velocity until it collides with another simulation particle or a solid surface. In the event of an intermolecular collision, an appropriate molecular collision model is employed to compute post-collision velocities, and in the event of gas-surface collisions, a suitable GSI model is adopted to calculate the molecular velocity after reflection. In this work, the internal energy exchange is modeled using the Larsen-Borgnakke scheme. Further, the intermolecular collision is handled using the variable hard sphere model and the no time counter-collision sampling technique. The simulation is always evaluated as an unsteady flow, and a steady result is obtained as the large-time state of unsteady simulation. After achieving a steady flow, the simulation particles in each cell are sampled for a sufficient duration to decrease statistical scattering. All macroscopic field quantities (such as mass density, velocity, and temperature) and surface quantities (such as surface pressure, shear stress, and heat flux) are calculated based on these time-averaged data. Results: Numerical results show that the distributions of gas pressure and mass flux are considerably affected by the GSI models. The lower the GSI accommodation coefficient, the higher the gas pressure and the larger the mass flux. Consequently, the GSI accommodation coefficients play a vital role in the compression factor and collection efficiency of the inlet. Furthermore, the decrease in the GSI accommodation coefficient from 1.0 to 0.2 leads to an increase in the compression factor and collection efficiency by a factor of 7 and 4, respectively. In addition, as the GSI accommodation coefficient decreases, the high-pressure region moves toward the ionization section, facilitating the ionization of neutral gas molecules. The following mechanism underlies this effect: after reflecting in a specular manner from the concave surface, the gas molecules congregate at the focus and enter the ionization section. Conclusions: To improve the inlet design of an ABEP system, a combination of the geometric design and surface-material design should be adopted. A concave compression section should be employed, and at the same time, the inlet surface should be smoothened to decrease the GSI accommodation coefficient.

Objective: The research on soil salinization is complicated, and the traditional literature review method struggles to grasp the development trend systematically due to its subjective nature. Hence, it becomes important and necessary to seek alternative methods to objectively summarize and analyze the existing research papers and reasonably guide the future development of this field. Methods: Based on the quantitative analysis of the Web of Science and China's national knowledge network databases related to soil salinization and saline-alkali land management and its utilization in the past 30 years, this paper proposes a quantitative review method for obtaining the research trend in this subject. This paper uses various software tools, including VOSviewer, Citespace, and SPSS, to analyze the number of publications, cooperation networks, and keywords within this field. Results: The results show a continuous increase in the research on soil salinization at home and abroad. The number of scholars participating in the study of soil salinization has also increased significantly. The primary stage of development was before 1999, the stage of stable development was from 2000 to 2011, and the stage of rapid development was from 2012 to the present. Despite significant progress, according to Kuhn's model, the field remains in the primary science stage, indicating ample room for development. English journals such as Agricultural Water Management and Science of the Total Environment and Chinese journals such as Soil Bulletin and Soil can be regarded as core journals in the field of soil salinization, with the highest number of publications and higher journal impact factors. China, the United States, India, Australia, and other countries (in order of the number of papers published by them) have made outstanding contributions to soil salinization research, with Chinese scholars leading in terms of the highest number of published papers and are the main force in the study of soil salinization. The cooperation network analysis shows the importance of research institutions and direct government agencies in promoting institutional cooperation. However, at present, most cooperations are limited to intra-institutional cooperation, but future efforts should focus on the positive impact of cross-context, cross-institutional, transnational, and interdisciplinary cooperations on leapfrog and diversified development of soil salinization research. Keyword burst detection shows that "biochar", "yield", "salt stress", "quality", "freeze-thaw cycle" and "water and salt transport" are the recent hot spots of concern. Additionally, keyword co-occurrence and trend analyses show the shift in research focus from irrigation and drainage and saline-alkali land improvement in the 1990s to soil-plant salt interaction mechanism, salt-tolerant plant cultivation, soil water and salt regulation mechanism, optimization control technology, research and development technology of saline-alkali soil amendment, large-scale applications, remote sensing monitoring of saline-alkali land, and salinization impact assessment. It represents the developmental trend of research on soil salinization and saline-alkali land management and its utilization from drainage improvement to comprehensive utilization. Conclusions: The research results of this paper quantitatively highlight the research hotspot and developmental trend of soil salinization and provide a reference for relevant researchers to grasp the developmental trends of the field and explore valuable new research directions.

Objective: Seismic damage and destruction of the stations, tunnels, and other structures considerably impair the functionality of the urban rail transit system. Current research on the system performance of the rail transit network primarily focuses on the scenarios of intentional attack and stochastic damage, which is dramatically different from the earthquake disaster scenarios. This paper proposed a quantitative framework to evaluate the seismic performance and resilience of rail transit networks. Methods: The seismic fragility model was used to calculate the failure probability of the primary structural elements, including stations, tunnels, and bridges of the rail transit system. The graphical model of the network was established using the Space L modeling method. This approach was used to depict the interdependency of system elements. The network performance was expressed by the network efficiency weighted by passenger flow between rail transit stations. The Monte Carlo simulation was used to assess the uncertainty of the earthquake damage state of structures and the post-earthquake recovery of the damaged elements. According to the network performance curves during the post-earthquake recovery process, the seismic resilience index and resilience loss of the rail transit network were quantitatively evaluated using the concept of resilience triangle. Considering the Beijing rail transit network, the effects of earthquake intensity, recovery strategy on network performance, and resilience indexes were investigated. Results: The results of the present analysis were as follows. (1) The resilience characteristics of rail transit networks under earthquakes, intentional attacks, and stochastic damage were different. The resilience index under earthquake damage was 0.936 3, whereas the resilience index under stochastic damage was 0.934 0. The resilience index under intentional attack was 0.863 4. (2) In the damage scenario corresponding to different earthquake intensities, the system resilience index calculated by the recovery sequence sorted by the dynamic importance of damaged elements were larger than that sorted by the static importance of damaged elements. Moreover, the damage scenario involving several damaged elements typically results in a larger difference between the resilience index calculated by the two recovery strategies. (3) Pre-earthquake enhancement measures to reduce the failure probability of crucial elements could effectively enhance the disaster resistance capacity of the network; however, their influence on improving the post-earthquake recovery capacity remained unclear. Conclusions: Based on the seismic fragility models of the primary structure of the rail transit network, the graphical model of the network, and the importance of ranking-based post-earthquake recovery of the damaged elements, this paper establishes a framework to quantitatively evaluate the seismic resilience of rail transit network by the passenger-weighted network efficiency. When evaluating network resilience and comparing antiseismic improvement measures, multiple indicators such as resilience index, resilience loss, and recovery duration should be comprehensively analyzed. This framework can provide a reference for the seismic performance evaluation of the urban rail transit network and help decision-makers in allocating maintenance resources to restore the operation function of the urban rail transit system in a timely and cost-effective manner during the recovery process.

Objective: The increasing maturity of large language model technology has facilitated its widespread application in downstream tasks across various vertical fields. Large language models have exhibited beneficial performance in text summarization tasks in general fields, such as news and art. However, the highly specific language style in the judicial field and the unique complexity of judicial documents in terms of structure and logic make it difficult for large language models to generate judicial document summaries. This study aims to combine prompt learning with large language models to explore their performance in summarizing judicial documents. Prompt templates containing structural information and judicial documents are used as inputs for fine-tuning large language models. As a result, large language models can generate judicial document summaries that adhere to judicial language styles and the structural and logical complexities of judicial documents. Methods: This study proposes a judicial document summary method that combines prompt learning and the Qwen large language model. Judicial document data are used as the input for fine-tuning a large language model using supervised fine-tuning technology to enhance its applicability in the judicial field. Simultaneously, prompt templates that incorporate structural information and role instructions are designed to optimize summary generation to more accurately reflect the structural characteristics and logical relationships of documents. According to the characteristics of the pretraining data format of the large language model, the fine-tuning data were constructed in the form of question-answer pairs. Results: The experimental results show that the proposed method improves the F1 of the baseline model by 21.44%, 28.50%, and 28.97% in ROUGE-1, ROUGE-2, and ROUGE-L, respectively, and exceeds all of the comparison models. The ablation experiment demonstrated that the summary generation method using prompt learning was superior to the method without prompt learning for all indicators, and the performance of summarization generated by the large language model utilizing prompt learning was significantly enhanced. The case demonstration reveals that after prompt learning is used to enhance the perception of structural information in the judgment document by the large language model, the judgment document summary generated by this model can better capture and retain key information in the judgment document. Moreover, the language style of this model is closer to that of a real judgment document summary, which further illustrates the effectiveness of the proposed method. Conclusions: This study integrates the structural information of a judgment document into the task of generating a judgment document summary using a large language model in the form of prompt templates. Prompt templates containing structural information are used to assist the large language model in summarization generation. Therefore, the model can focus on the key information in the judgment document and capture deeper semantic logical relationships. The results demonstrate that after fine-tuning the large language model with judicial document data and introducing structural information, the model demonstrated excellent performance and great application potential in the judicial document summary task. The proposed method can effectively enhance the capability of a large language model in the field of judicial document summaries.

Objective: Ontology structure has been proved to be particularly important in the construction and organization of a knowledge graph (KG). A comprehensive method for the modeling, updating, and application of KG with the guidance of domain ontology needs to be explored. In view of the common knowledge gap in engineering safety management, this paper aims to propose an ontology-based framework to achieve domain knowledge modeling and updating. Using the highway engineering field as an example, this paper demonstrates how safety knowledge can be automatically extracted from industry-standard text data to facilitate the construction of a domain KG. Subsequently, the safety management scenarios are developed based on the building information model (BIM), and the auxiliary role of intelligent knowledge in safety management is demonstrated to verify the effectiveness of the engineering application of the developed KG. Methods: This paper used the ontology-guided domain knowledge extraction method to construct the domain KG and proposed a knowledge network-guided method to update the ontology. Specifically, a layered knowledge system with multiple dimensions was summarized as the ontology layer based on the management approach and the established standard specifications within highway engineering. Following the guidance of the ontology layer, a structured knowledge network acting as the data layer was extracted from massive text materials by developing a series of knowledge extraction procedures. Consequently, a knowledge-flowing method from the data layer to the ontology layer was proposed. Three categories of methods based on the essence and composition of entities and the clustering of the entity's core words were summarized to realize the automatic updating of the ontology layer. Finally, combined with the developed highway safety information retrieval and application system, this paper demonstrated the organization and application of the constructed domain KG, thus verifying the effect of introducing ontology in the organization and deployment of knowledge. Results: The developed ontology of highway engineering safety knowledge was featured as a layered knowledge system with seven levels and 390 nodes connected with ~300 000 valid entity nodes in the data layer, facilitating the creation and integration of the KG's logical structures. The proposed method for updating the domain ontology aided by over 1 000 technical terms was demonstrated to be effective, with an increment of 51.5% in the expansion of the nodes to the ontology. The designed method of linking the ontology-guided domain KG with the BIM was validated for its feasibility through practical implementation within a real highway engineering safety management system, displaying the positive impact of ontology's guidance in the organization and expansion of knowledge. Conclusions: This paper concentrates on the domain of highway engineering safety and presents a comprehensive paradigm for constructing, updating, and applying a domain KG to demonstrate methodological innovation in ontology updating. The results extend the application scope and technical approaches of KG technology, thereby enhancing information technology level in engineering safety management. Moreover, the findings of this research can be used for BIM evaluation and safety guidance in highway engineering construction projects, thereby advancing the level of information technology in construction safety management.

Objective: Threat assessment of targets serves as a critical reference for commanders in wartime decision-making. With the rapid development of unmanned systems and smart technologies, the future of warfare is progressing toward unmanned, multi-domain, and clustered operations. However, existing studies on target threat assessment fall short of effectively satisfying these demands of future warfare, demonstrating three main issues: 1) Majority of combat scenarios focus on singular settings, such as maritime air defense, air-to-air combat, and ground-based air defense, with scant research on multi-domain operations (land, low-altitude, and electromagnetic environments). 2) Research is mainly concentrated on individual entities or cluster targets, such as fighter aircraft, unmanned aerial vehicle swarms, and unmanned surface vessels, with inadequate investigation of clustered equipment integrating manned/unmanned ground combat vehicles and low-altitude manned/unmanned aircraft. 3) Current methodologies predominantly consider the state and characteristics of Blue Force targets, ignoring the influence of dynamic changes in Red Force equipment on the weighting of threat indicators for Blue Force targets. Methods: To deal with these problems, we proposed a dynamic assessment method for threats to clustered targets in low-altitude, multi-domain battlefields based on hesitant fuzzy sets. First, we explored the laws governing low-altitude, multi-domain battlefields and the operational characteristics of clustered equipment that involves manned/unmanned air and ground elements. We analyzed five major influencing factors in the threat assessment of cluster targets, namely, operational cluster type, urgency, comprehensive strike capability, intelligent collaborative capability, and importance of the attack area, and determine an indicator system for threat assessment of clustered targets. Afterward, leveraging the Weber-Fechner law, we explored the relationship between changes in the Red Force's situation and the psychological pressure experienced by commanders and proposed a Weber-Fechner law-based weight determination method, which adjusted the weight values of the Red Force's comprehensive strike capability and the Blue Force's air power strike capability in conjunction with variations in the damage rate of the Red force's air defense capability. Finally, by combining the variable weight method under a hesitant fuzzy environment, a dynamic assessment model based on hesitant fuzzy sets for threats to cluster targets in low-altitude, multi-domain battlefields was constructed. Results: In a simulation, when the Red Force's air defense system sustains serious damage, the threat posed by the Blue Force's air power intensifies significantly. By utilizing the Weber-Fechner law-based weight adjustment method, the weight determination becomes more scientifically reasonable, effectively and promptly reflecting the psychological changes encountered by commanders when faced with the stimulation of the battlefield situation and reducing the subjectivity and arbitrariness related to weight optimization adjustments. Comparative analysis of the threat assessment results under constant and variable weights demonstrates that cluster targets with air power superiority exhibit more sensitive and timely adjustments in threat assessment results under variable weight conditions with a higher level of consistency. Conclusions: These results further confirm the accuracy and effectiveness of the model, providing commanders with feasible and reliable decision support.

Objective: Emergency relief supplies are crucial for dealing with disasters, and their reasonable and timely distribution relates to people's health and safety. Emergency relief supplies at rescue centers are limited and cannot meet the emergency needs of all affected areas simultaneously. Post-disaster emergency relief supplies face double challenges in this regard due to short supplies and limited transportation capability, resulting in the needs for medical rescue and materials of a disaster area in a short period. To develop a scientific and efficient post-disaster emergency response, we studied the constrained vehicle routing of emergency relief supplies based on demand urgency. Restrictions in traditional path planning, such as single objective, single depot, single distribution, undifferentiated supply, and closed scheduling, were considered. Methods: The analytic hierarchical process was applied to measure the demand urgency index from personnel, facilities, and disaster resistance, considering the overall efficiency and key disposal. Furthermore, this study had multiple objectives, including the following: minimization of deprivation cost and response time, and maximization of demand satisfaction rate in the emergency rescue process. A constrained model of emergency vehicle routing was constructed, and a two-stage genetic algorithm was designed to deal with comprehensive distribution conditions, such as open scheduling, soft time windows, and demand splitting. The effectiveness and feasibility of the model and algorithm were verified using examples. Results: The results revealed that the model effectively coped with the material distribution problem resulting from scarce transportation capacity and various degrees of disaster. The splitting strategy and open scheduling of vehicles guaranteed multiple services at disaster sites and optimal route combination. Moreover, relief progress in disaster sites (splitting demand, batch distribution, and service time) and vehicle dispatch schedules (distribution order, work duration, and resupply depot) were generated. During the planning period, the system loss was reduced by 40.3 %, and a 99.4 % material demand was obtained. When disaster derivation caused changes in road conditions, fluctuation parameters were inputted into the model. The model and algorithm adjusted the scheme with a low risk of service failure, and the adjusted scheme reduced the demand and supply by 1.5 % in the decision period. Conclusions: Constrained route planning is implemented for flexible distribution conditions, such as demand splitting, soft time windows, and open scheduling, based on the dynamic change characteristics of demand and supply during sudden natural disasters. This study considers the demand urgency of key disaster areas and the efficiency of global relief to accommodate unexpected road events and maximize resource availability. With the circulation of distribution vehicles, the needs of disaster sites are gradually met within the decision-making cycle, which provides full play to the time utility of emergency relief supplies and transportation resources. The proposed model can form scientific and reasonable material distribution and vehicle scheduling schemes and evaluate the workload of each rescue center and vehicle to deploy work in advance, providing a theoretical basis and a decision-making reference for vehicle route planning of emergency relief supplies.

Methods: To more clearly describe the specific movements of each joint, the local POE method was introduced. For ease of analysis, the structure of the robot's passive arms was simplified using screw theory. A kinematic model for the Delta robot was established using the local POE method. The error model of the robot was obtained through the differential mapping of the exponential product. Based on the derived error model, error sources were subdivided into three major categories: structural errors, actuation angle errors, and spherical joint clearance errors. An in-depth analysis was conducted on how each error source affects the end-effector positioning accuracy of the robot when it moves along the X, Y, and Z directions. A Delta robot with active arm lengths of 400 mm and passive arm lengths of 950 mm was selected as the subject for simulation analysis in MATLAB. The square root of the sum of squared errors in the X, Y, and Z directions was used as a composite error to serve as an evaluation criterion. Results: The simulation results showed that assuming all error sources have a magnitude of 0.100 units (length unit being mm; angular unit being degrees), actuation angle errors had the most significant impact on the end-effector positioning accuracy of the Delta parallel robot, causing a composite error ranging from 1.500 to 2.000 mm. Spherical joint clearance errors caused a composite error of 0.340 mm in the robot. Structural errors exhibited a relatively stable composite error fluctuating around 0.100 mm, with a variation range of approximately 0.010 mm, which can be considered a constant value. Comprehensive analysis indicated that length errors in the active and passive arms significantly influenced end-effector positioning accuracy, with the induced error fluctuations notably larger than those from other sources. Additionally, when the magnitudes of error sources were 0.025 mm, 0.050 mm, 0.075 mm, and 0.100 mm, their impacts on robot positioning accuracy increased proportionally. Conclusions: The Delta robot error analysis model based on screw theory and utilizing the local POE method offers a more intuitive and comprehensive approach to analyzing the impact of major error sources on positioning accuracy compared to traditional error modeling methods. This approach effectively avoids issues of singularity and incompleteness. It provides theoretical reference for error modeling analysis of other parallel mechanisms. Through the assessment of the influence of each error source presented in this paper, during subsequent error compensation phases, more precise corrections can be made to the significantly impactful actuation angle errors, thereby effectively improving the efficiency and effectiveness of overall error compensation.

[Objective] An effective causal analysis of accidents is essential for learning from and preventing coal mine accidents. Manual analysis of accidents is strongly influenced by the subjectivity of the personnel involved and becomes inefficient for analyses involving large volumes of accident and risk text data. Although considerable research has been conducted in the area of accident text mining, most studies directly apply data mining techniques to extract accident information and factors from texts without considering accident causation theories. This approach leads to results that lack systematic and logical coherence. [Methods] To address the aforementioned issues, this paper proposes a method for the intelligent identification of accident causes in the coal mining sector. This method integrates entity recognition, semantic dependency analysis, text classification, and the accident causation “2-4” model (24Model). Specific implementation steps for this method are also provided. Accident causation theory is crucial for ensuring the effectiveness and scientific validity of accident analysis. This paper introduces the 24Model as a theoretical basis for accident cause identification, and the advantages of the model in the intelligent analysis of accident causes are highlighted. Entity recognition technology is employed to identify key entity information in accident texts, including information on personnel, organizational structures, accidents, abnormal characteristics, values, safety management, building facilities, environments, equipment and materials, safety policies, procedural documents, and operational processes. To effectively identify this information, this paper integrates the bidirectional encoder representations from transformers(BERT)-bidirectional long short-term memory(BiLSTM)-conditional random fields(CRF) model and trains the combined model using 660 accident texts. This paper utilizes semantic dependency analysis technology to identify the semantic relationships among entity information. Text representation patterns were extracted according to the definitions of unsafe individual actions and organizational factors by the 24Model, and these definitions were used to determine the types of accident causes. This paper utilizes a text classification method to develop a model for identifying individual capability-related causes, and the focus is on five aspects: knowledge, awareness, habits, and psychological and physiological factors. The text classification model was based on BERT. This paper evaluates the accuracy of the proposed method in identifying accident causes by comparing both the entity recognition model and the text classification model with similar models. Test cases were selected, and the results of accident cause analysis via the proposed method were compared with those from manual analysis. Additionally, this paper develops an application based on the proposed method to facilitate the analysis and learning of onsite accident cases by employees of coal mining enterprises. [Results] This research results showed that the precision rates of the trained entity recognition model and the text classification model reached 95.42% and 96.11%, respectively. Additionally, the accuracy of the accident cause identification method, when combined with semantic dependency analysis, reached 73.09%. [Conclusions] The contributions of this paper are as follows: (1) Integration of the definition and concept of the 24Model and the automatic identification of unsafe behaviors according to the 24Model. This approach helps avoid the strong subjectivity and inconsistency often present in accident analysis conducted by different personnel. (2) Further identification of the actors of actions, operational procedures, materials, and equipment. (3) Fusion of multimodel algorithms to identify the causes of accidents, allowing for the rapid analysis of a large number of accidents. (4) Facilitating the application of accident causation theory in coal mining enterprises, enhancing the effectiveness of accident case analysis and learning, thereby achieving the objective of preventing related accidents.

Objective: The structural integrity of bridges is a critical concern as infrastructure ages, necessitating the development of reliable methods for detecting potential failures. Among these, the identification of small target cracks is particularly important, as these cracks often grow undetected until they result in severe damage. Traditional inspection methods, such as manual visual inspections, are hindered by their labor-intensive nature and susceptibility to human error, often resulting in the oversight of small but significant defects. Recent advancements in computer vision and deep learning technologies offer new opportunities to improve the accuracy and efficiency of bridge inspections. This study introduces an innovative approach for detecting small target cracks in bridge structures by employing an enhanced version of the You Only Look Once (YOLOv8) object detection model, a widely recognized algorithm known for its rapid processing capabilities and high detection accuracy. The enhanced YOLOv8 model is tailored to detect small-scale cracks on bridge surfaces that may not be easily identifiable by traditional inspection methods or earlier versions of computer vision models. Methods: The proposed algorithm modifies the standard YOLOv8 model to address the specific challenges associated with detecting small cracks on bridge surfaces. A key modification is the integration of efficient vision transformer (EfficientViT) into the backbone of the YOLOv8 model. EfficientViT is an advanced transformer-based architecture that reduces redundant parameters and optimizes the extraction of local features from high-resolution images, enabling more precise detection of subtle crack features. This enhancement is crucial, as small cracks often exhibit low contrast against their background and may be easily overlooked by less sophisticated models. In addition to EfficientViT, the proposed algorithm also incorporates large selective kernel network (LSKNet) within the C2f module of YOLOv8. LSKNet employs a dynamic kernel selection mechanism that allows the model to adaptively adjust the size of the convolutional kernels based on the input features, making it highly suitable for detecting cracks of varying sizes, orientations, and morphological characteristics. This adaptability ensures that the model can detect small cracks, regardless of their form. Furthermore, the model uses bidirectional feature pyramid network (BiFPN) to merge feature maps at different scales. Traditional models struggle with detecting small targets due to the loss of critical information during downsampling operations. BiFPN mitigates this issue by preserving high-resolution feature maps across multiple layers, enhancing the model's ability to detect small cracks that would otherwise be missed. The combined effect of these modifications improves the accuracy of small target crack detection while maintaining computational efficiency. Results: The effectiveness of the proposed model was validated using a dataset of crack images from a specific bridge, captured by unmanned aerial vehicles (UAVs). UAVs provided detailed images from areas that were often difficult or dangerous to access using traditional inspection methods. The experimental results demonstrated that the enhanced YOLOv8 model significantly outperformed the original version in terms of key performance metrics. Specifically, the modified model achieved improvements of 3.7%, 3.5%, 3.5%, 3.9%, and 7.4% in terms of the detection precision, recall, F₁ score, mAP50, and mAP50-95, respectively. These results indicated a substantial improvement in the model's ability to detect small cracks that often had low contrast and irregular shapes, which were typical characteristics of cracks on bridge surfaces. Furthermore, compared to conventional methods, the proposed model was able to detect cracks with higher precision and fewer false positives, making it a promising tool for improving the efficiency of bridge inspections. Conclusions: In conclusion, the improved YOLOv8 algorithm introduced in this study represents a significant advancement in the detection of small target cracks in bridge structures. The modifications made to the original YOLOv8 model, including the integration of EfficientViT, LSKNet, and BiFPN, result in a more accurate and computationally efficient model for crack detection. This approach offers a practical and scalable solution for the widespread application of bridge health monitoring, particularly in areas that are difficult to inspect using traditional methods. By leveraging advanced surface data processing techniques, this research contributes to the development of modern methods for assessing the health of bridge structures, ultimately helping to ensure the safety and longevity of infrastructure systems.

Objective: The two-dimensional (2D) hydrodynamic model can simulate the process of flood inundation and evolution. This model is widely used in flood forecasting. The number of grids and time steps considerably affect the computational efficiency of this model. Various methods have been proposed to improve the computational efficiency of the 2D hydrodynamic model. Some researchers use the dynamic grid strategy to calculate only effective grid cells to reduce the impact of the increased number of grids. Others use the local time step (LTS) technology to decrease the time consumption caused by small time steps. Whether the efficiency of model computation can be improved by combining the advantages of the two strategies requires further research. Methods: Herein, a hybrid algorithm that combines the dynamic grid strategy and LTS technology is proposed to further improve the model performance based on the self-developed flood simulation model, HydroMPM. Technically, the grid cells that actually assist in the flux calculation are first selected as effective cells. Then, the LTS technology is applied to these cells to further optimize the flux calculation and update strategy. The calculation accuracy and efficiency of the hybrid algorithm are compared and analyzed using an ideal dam break case and a typical flood simulation scenario in the Nangang River basin, Guangdong Province, China. Results: The dynamic grid strategy can accelerate model computation by computing only the effective grid cells. However, the effective cells actually contain all computation grids when the computation area is completely submerged. In this case, the dynamic grid strategy may lead to a high computation amount and low computation efficiency owing to the dynamic update mechanism on all grid cells. The LTS technology can improve the average time step by hierarchically updating the grid cells. However, the performance of this technology barely depends on the difference in grid scale and flow velocity distribution. The urban flood process often has a scattered distribution of the local inundation area, which is suitable for the application of the dynamic grid strategy. At the same time, local mesh refinement is also required in urban flood simulation. This refinement enables the model to better describe the topographic variation in local waterlogging-vulnerable areas. However, it also leads to a large difference in the maximum time step allowed by different grid scales, which requires the application of the LTS technology. By combining the two strategies, the hybrid dynamic grid and LTS technology can further enhance the model performance. In the ideal dam break case, the hybrid algorithm can reduce computation time by 13.1%-64.8%. In the practical application case of the Nangang River basin, the calculation time can be saved by approximately 60%. Conclusions: The hybrid algorithm successfully combines the advantages of the original dynamic grid strategy and LTS technology to further accelerate the efficiency of the model computation. However, the performance of this hybrid algorithm varies depending on the application scenarios. If the 2D hydrodynamic model is used to simulate only river floods, estuaries, and other areas, the effect of the dynamic grid strategy may not be obvious. The LTS technology may have a general effect when the grid distribution is uniform and the flow state is stable. The hybrid algorithm can combine the advantages of the two abovementioned algorithms. However, this hybrid algorithm has the inherent shortcomings of the two algorithms. In practical applications, a suitable algorithm should be chosen depending on the specific application scenario to achieve good simulation effect and computing efficiency.

Objective: In the expansive field of geology, where anisotropic fractures intricately pattern rocks, this study focuses on unraveling the nuanced evolution mechanisms of microscopic cracks within rocks featuring a solitary joint. Rooted in the jointed rock mass of the Xianglushan Tunnel in the Dianzhong Diversion Project, China, the research aims to discern the profound impact of single joints on a broad spectrum of macroscopic mechanical parameters and failure characteristics. This exploration seeks to deepen our understanding of the intricate interplay between micro-mechanical phenomena and the broader geological context, contributing valuable insights to the field of rock mechanics. Methods: In this pioneering study, the methodology hinges on leveraging the advanced two-dimensional particle flow code (PFC2D) to meticulously orchestrate uniaxial compression simulation tests. The experimental scope spans both pristine rock specimens and those featuring a distinct single joint. The crux of the analysis entails a detailed exploration into the repercussions of joint length and inclination on a diverse array of macroscopic mechanical parameters and failure characteristics.To dissect the intricate relationships between joint attributes and the mechanical response of the rock mass, the study employs numerical experiments. These simulations, akin to a virtual laboratory, diligently replicate the dynamic response of the rock mass under varying joint conditions. The computational prowess of PFC2D ensures a high-fidelity representation, unraveling the nuanced interplay between joint characteristics and macroscopic mechanical behaviors.The numerical experiments extend beyond the confines of traditional physical testing, enabling a systematic investigation across a spectrum of joint conditions. This not only enhances the efficiency of the study but also broadens the horizons of exploration, providing insights into diverse joint scenarios that might pose challenges in a laboratory setting. Results: The results indicated that for rocks with a single joint: (1) smaller joint inclinations and larger lengths corresponded to decreased uniaxial compressive strength, peak strain, and elastic modulus. (2) Longer joints exhibited increased sensitivity of joint inclination to peak stress, peak strain, and elastic modulus. (3) Specimens predominantly underwent tensile failure, with a sequence of crack initiation: wing cracks, shear cracks, secondary shear cracks, and far-field cracks. (4) As the joint inclination increased, the crack initiation location shifted from the middle to the tip of the joint, and the crack initiation direction changed from perpendicular to the joint strike to parallel. (5) Longer joints resulted in fewer primary tensile cracks, simpler crack types, earlier initiation of wing cracks, and delayed initiation of shear cracks. Conclusions: This groundbreaking research represents a significant leap forward in unraveling the micro-mechanical intricacies inherent in single-joint rocks and their profound implications on macroscopic mechanical parameters and failure characteristics. The acquired insights not only substantively contribute to the academic discourse in the field of geomechanics but also hold practical implications for the assessment and prediction of the mechanical behavior of jointed rock masses in engineering applications. The findings serve as a cornerstone, providing a robust foundation for future research endeavors in the dynamic realm of rock mechanics. The practical implications extend beyond theoretical boundaries, offering valuable guidance for engineers and practitioners engaged in the design and evaluation of structures within jointed rock formations, thereby bridging the gap between theoretical understanding and real-world applications.

Significance: Aerospace vehicles have undergone significant modifications in terms of aerodynamic shape, flight speed, flight environment, and flight duration compared with conventional flight vehicles. They must withstand harsh aerodynamic thermal environments for long durations and maintain a sharp leading-edge shape with a high lift-to-drag ratio, imposing extremely stringent requirements on the temperature resistance, durability, structural efficiency, and reliability of the thermal protection system. Traditional thermal protection depends largely on passive methods such as heat insulation, heat sink, and radiation heat dissipation. Although the thermal protection performance of related technologies has improved, which is restricted by several constraints, such as ensuring that the prototype is safe under harsh conditions of extremely high heat flux and ultrahigh temperature along with structural stability, long-term operation, light-weight nature, and repeatability. Thus, a new active thermal protection technology is necessary. In this context, transpiration cooling technology offers the advantage of high thermal efficiency without requiring any changes in the prototype of a vehicle. It has been widely considered a potential active thermal protection technology. However, when transpiration cooling is used for thermal protection of a flight vehicle, some challenges related to the complexity of the system, a mismatch between coolant supply and demand, unstable control of the operation, and development of a high-precision prediction model etc., arise. Progress: Research on transpiration cooling primarily focused on quick evaluation of performance, numerical simulation of flow and heat transfer, evaluation of cooling mechanism performance, development of optimal control algorithm for efficiency, and optimization of structure form and yielded beneficial results. However, several fundamental scientific issues needed to be urgently addressed to fully realize the engineering application of this technology in aerospace vehicles. In the context of numerical simulation, the accuracy and adaptability of the heat and mass transfer model should be improved. Most existing studies had mathematically described and solved the physical process of heat and mass transfer in porous media at the macroscale. But some parameters related to specific phase change heat and mass transfer (such as evaporation/condensation coefficient and fluid-solid convection heat transfer coefficient) that affect the model's accuracy must be modified through experiments, and the adaptation was partially successful. Most existing models assumed that the temperature of porous media, liquid phases, and gas phases were equal. Although a few models explored the nonequilibrium effect between porous media and fluids, they did not consider the nonequilibrium effect between gas and liquid phases. There were few flight experiments in the research and a large gap between the ground experimental test and practical use conditions. Furthermore, extreme effects related to high-temperature, real, and rarefied gases and shock wave/boundary layer interference during high-speed flight could not be effectively reproduced on the ground. Moreover, there was a lack of experimental data that could be used to verify the accuracy of the heat and mass transfer model. The experimental test method was relatively simple, and the flow and heat transfer process of the liquid in the porous medium could not be obtained. It was challenging to effectively obtain the boundary layer flow law of the liquid when it entered the high-speed mainstream flow from the porous medium. In terms of control strategy, the present research on transpiration cooling control systems lacked a transient simplified mathematical model that could be quickly established, particularly for liquid phase change transpiration cooling with the multiphase flow and phase change process. Simultaneously, there were few transpiration cooling control systems with practical engineering values based on modern control theory, which made it difficult to achieve optimal performance in practical engineering applications. Some adaptive and self-driven transpiration cooling systems had been proposed as new forms of transpiration cooling structures; however, they were still at the mechanism verification stage, and the engineering application effect needed to be verified. Conclusions and Prospects: Follow-up research will focus on the micro/mesoscale fine numerical calculation model, advanced visual experimental testing methods, rapid response-precise control strategies, self-driven and adaptive structural engineering systems, and combined active and passive thermal protection.

Objective: Single-photon emission computed tomography (SPECT) is still considered a nonquantitative imaging modality because it cannot perform attenuation correction. To acquire a quantitative SPECT image, Beijing Novel Medical Equipment Co., Ltd. developed a domestic SPECT/CT system, Insight NM/CT Pro. It was equipped with dual digital detectors and enabled various scanning geometries. In addition to better spatial resolution, the system can correct attenuation effects in SPECT and realize quantitative reconstruction using CT. In this study, we evaluated the imaging accuracy of this domestic SPECT/CT system and compared it with advanced abroad systems. Methods: To obtain quantitatively correct results, physical effects, including collimator blurring, object attenuation, scatter, and radionuclide decay, were modeled and corrected in the reconstruction algorithm. Moreover, a large cylindrical phantom was employed to obtain the calibration factor for the system so that we could convert the reconstruction result to a quantitative image in terms of Bq/mL. Then, the performance of CT-based attenuation correction was tested according to testing method for SPECT imaging based on CT-attenuation correction (YY/T 1546—2017). A cylinder phantom with three cylindrical inserts corresponding to air, nonradioactive water, and bone was filled with a radioactive solution for imaging. The biases inside the air, water, and bone regions of the reconstructed image were calculated to evaluate the performance of CT-based attenuation correction. In addition, the quantitative accuracy of the equipment was tested using the Notional Electrical Manufactures Association (NEMA) torso phantom according to performance measurements of Gamma cameras (NEMA NU 1—2018). Six fillable spheres with different diameters were set as targets for recovery evaluation, and the target-to-background concentration ratio was approximately 8∶1. A large volume of interest (VOI) was placed in the background region to calculate the quantitative bias of the reconstruction. VOIs with the same size of six spheres were drawn based on the registered CT to evaluate recovery coefficients of different sizes. Radionuclide technetium-99m and a low-energy, high-resolution collimator with Insight NM/CT Pro were used for all these tests, and the evaluation results were compared with those of the GE Discovery NM/CT 670. Results: For attenuation accuracy testing, the biases inside the air, water, and bone regions of Insight NM/CT Pro were 7.84%, 8.38%, and 4.66%, respectively, whereas the corresponding biases in GE Discovery NM/CT 670 were 16.64%, 18.01%, and 11.02%, respectively. For quantitative accuracy testing, in GE discovery NM/CT 670, the quantitative recovery coefficients of the hot spheres with diameters of 13, 17, 22, and 28 mm were 31.04%, 51.36%, 59.91%, and 66.39%, respectively, and the background concentration bias was 10.84%. Insight NM/CT Pro achieved higher recovery coefficients and lower bias. The recovery coefficients were 38.22%, 50.98%, 66.55%, and 71.32% for 13, 17, 22, and 28 mm hot rods, respectively, and the bias in the background region was 7.95%. Conclusions: Phantom studies have demonstrated that the domestic Insight NM/CT Pro imaging system can obtain smaller biases after CT attenuation correction and achieve quantitative images with high accuracy. The reliability of the Insight NM/CT Pro system for quantitative imaging is validated in this study, and its performance is comparable to that of the advanced GE Discovery NM/CT 670 imaging system.

Objective: Risk quantification is crucial in risk assessment of accidents or disasters. This study aims to investigate the risk quantification method utilized in over-temperature faults in electrical circuits. The existing technologies of the abovementioned method are summarized in electrical and fire signals of disaster early warning. Thus, a new method based on the fire big data is proposed. Methods: Different frequency electrical parameters are collected by the detector arranged at the front end and transmitted in real time to the fire big data to mine the influencing factors and changing patterns of electrical circuit temperature based on the deep-learning method. Subsequently, the probability distribution of temperature is determined statically, and risk is described by comparison of prediction temperature in different cumulative probabilities with actual temperature. To predict the electrical circuit temperature, a recurrent neural network (RNN) is utilized to model temperature prediction. The input parameters are voltage, current, temperature, and residual current. Among the parameters, there are two data sources for the model: one is real electrical fire data, 6 min^-1, used to learn the periodic law of temperature increase in electrical circuits of RNN for low-frequency data (LF-RNN), and the other is experimental data based on simulated fault of the temperature increase in electrical circuits. This experiment is implemented in three-phase resistive electrical circuits. Exceeding rated current is utilized to produce temperature rising. Meanwhile, electrical parameters are collected to study the law of temperature oscillation of RNN for high-frequency data (HF-RNN). Among these electrical parameters, the sampling frequency of voltage, current, and residual current is 50 kHz, but 1 Hz for temperature exceptionally. The optimization method, hyperparameter traversal, aims to minimize the loss function and root mean square error; thus, temperature prediction in electrical circuits is preliminarily applied. To increase the accuracy of the prediction model and elucidate the relationship between fire risk and prediction result, a temperature probability prediction model is established based on its second-order residual normal distribution. The error and its reducing methods are analyzed, and the relationship between prediction error and temperature mutation is determined. Results: The results demonstrated that temperature mutation within three window lengths had a remarkable linear correction effect with temperature prediction error; moreover, the second-order residual approximately followed a normal distribution. The upper and lower limited of temperature prediction confidence intervals with different significant levels (α=0.02, 0.04, 0.06, …, 0.98) can be computed by interval estimation, which had a one-to-one correspondence with temperature prediction accumulate probability $ \left(1-\frac{\alpha}{2}\right)$ and $ \frac{\alpha}{2}$. The results revealed that the cumulative distribution probability 1% prediction curve and 99% prediction curve appeared to have a fine coverage effect on the actual temperature. With the aim of measuring temperature prediction probability distribution with electrical fire risk, the concept of "early-warning quantile" similar to cumulative distribution probability, was proposed. The ability to predict temperature was established using different "early-warning quantile curves" and confirmed through 2 943 sets of real electrical fire scene data. The results demonstrated that early-warning quantiles in the range of 10%-30% could overlap the majority of the actual temperature data, and the higher the quantile of the curve was, the higher the frequency of overestimating the temperature was. Conclusions: To summarize, when the temperature in electrical circuits suddenly increases, there is a substantial upward trend in the early-warning quantile of the actual temperature. Thus, the use of LF-RNN and HF-RNN can timely and accurately predict the temperature probability distribution to characterize fire risks in electrical circuits so that early dynamic perception of fire risk is realized.

Objective: In recent years, a large number of nonconvex, highly nonlinear, multimodal, and multivariable complex optimization problems have emerged in scientific and engineering technology design due to the continuous development of science and technology. Owing to their advantages such as simple programming, flexible operation, and efficient optimization, intelligent optimization algorithms have become research hotspots to address diverse complex optimization problems in engineering applications. They have been successfully used to solve practical problems such as neural networks, resource allocation, and target tracking. In this research, multiple strategies were developed to improve the existing monarch optimization algorithm to address its shortcomings, such as slow convergence speed, low optimization accuracy, and ease of falling into local extremum. Methods: First, the forward normal cloud generator is used to perform nonlinear cloud operation on the parent monarch butterfly, increasing the number of candidate solutions and improving the local development ability of the algorithm. Subsequently, an opposition-based learning strategy based on convex lens imaging is used to the current optimal individual which is generated by normal cloud generator to generate new individuals and improve the convergence accuracy and speed of the algorithm. Finally, adaptive strategies are incorporated into the adjustment operator to diversify the population. Results: Several experiments were performed on benchmark functions to verify the performance of the algorithm: (1) Different strategies proposed were analyzed using ablation experiments to verify their effectiveness. The results revealed that the proposed strategies can effectively improve the algorithm's performance. (2) The improved algorithm was compared with other swarm intelligent optimization algorithms, and the results revealed that the improved algorithm can achieve the best results on most test functions. (3) The improved algorithm was also compared with other improved versions of monarch optimization algorithm, and the results revealed that the improved algorithm exhibited more advantages such as fast convergence speed and high convergence precision. (4) The Wilcoxon rank sum test and Friedman test were used to verify the performance of the proposed algorithm. The results revealed that the improved algorithm is superior to other algorithms. Conclusions: The optimization and comparison results of the pressure vessel design and welded beam design in engineering applications further verified the superiority of the improved algorithm in addressing real-world engineering problems.

Objective: This study proposed a fully coupled computational fluid dynamics-discrete element method (CFD-DEM) model based on a diffusion averaging algorithm for the hydraulic transport of dense particles, integrally considering particle--liquid interphase force and complex particle-turbulence interaction. The proposed model overcame the limitation that fluid mesh needs to be several times the size of the particles in the traditional CFD-DEM model. Moreover, experimental and numerical studies were conducted mainly on horizontal and vertical pipes, and few were conducted on inclined pipes. Methods: Calculation of particle volume fraction was divided into two steps. First, each particle was randomly and uniformly divided into several feature points, and the initial value of the particle volume fraction was calculated based on the number of feature points occupied in each mesh. Subsequently, a diffusion-based averaging method was employed to solve the particle volume fraction with the initial field and no-flux condition on all physical boundaries in the computational domain. Furthermore, the source terms were added to the k-ε turbulence model to account for the modulation of the turbulence from particles, and the discrete random walk model was used to calculate the stochastic effect of turbulence on particle motion. A drag force considering porosity modification was applied to the two-phase flow through densely packed particle beds. Other particle-liquid forces and particle torques caused by the fluid were also included in the model. The fully coupled CFD-DEM model predicted the hydraulic conveying of dense particles in the pipeline system well. Moreover, this model was used to investigate the effects of pipe inclination on the hydraulic transport of coarse particles (2 mm), including the effects on the spatial distribution of particles, axial velocity of each phase, fluid turbulent kinetic energy, and pressure drop. Results: The results are summarized as follows: 1) The spatial distribution of particles gradually transformed from a relatively densely packed distribution at the bottom of the horizontal pipe to a nearly uniform distribution in the vertical pipe with increasing inclination angle. The distributions of axial liquid velocity and turbulent kinetic energy along the vertical direction were gradually asymmetric and then returned to symmetry, reaching the maximum degree of asymmetry at 60°. 2) In the inclined pipes, the axial velocity of particles was lower and higher at the bottom and top of the pipe, respectively. Meanwhile, the axial velocity of the particles in the vertical pipe was parabolically distributed, with higher velocity at the center of the pipe and lower velocity near the wall. 3) The number of collisions between particles and between particles and walls increased slightly and then decreased rapidly with increasing inclination angle. 4) Moreover, pressure drop in the two-phase flow initially increased and then decreased with increasing inclination angle, reaching the maximum at 60°. Conclusions: This study demonstrates that the inclination angle significantly affects the distributions of particles, the number of collisions between particles and between particles and walls, liquid turbulent kinetic energy, and pressure drop. A small or large inclined angle is suggested for the hydraulic transport of particles, and a 60° inclined pipe should be avoided to reduce energy consumption.

Objective: Evacuation simulation is an important method for studying tunnel safety. Previous studies on highway tunnel evacuation simulation have assumed a uniform distribution of various vehicle types across each lane. However, this overlooks the potential impact of differences in vehicle type distribution across lanes (DVTDLs) and random vehicle distribution on evacuation time. In practice, different vehicle types tend to gravitate toward specific lanes, and the volumes as well as personnel loads vary significantly among these vehicles. We propose an algorithm for the random arrangement of vehicles within a tunnel, considering DVTDLs and random vehicle distribution. Concurrently, we develop a system with randomly arranging vehicle types (RAVT). Methods: Our proposed method randomizes the arrangement of vehicles using parameters such as the spatial coordinates of the numerical model of the tunnel, vehicle sizes, vehicle model ratios, and vehicle type distribution differences. We apply this algorithmic process to develop the RAVT system. This system, based on HTML5, CSS, and JavaScript, can quickly calculate the parameters of vehicle arrangements and personnel loads and automatically generate the corresponding fire dynamics simulator (FDS) codes. This feature enables efficient and accurate vehicle arrangement in the numerical model of fire and evacuation of road tunnels. The FDS codes serve as a tool for modeling fires and can be directly imported into Pyrosim (a graphical tool integrated with FDS) and Pathfinder (a graphical tool for evacuation simulation). Using RAVT and Pathfinder, we established 12 scenarios featuring two- and three-lane tunnels as the subjects of our research. A total of 240 simulations were conducted. Our analysis focused on the effects of DVTDLs and random vehicle distribution on evacuation time under various conditions, ranging from extremely congested (vehicle spacing of 1.5 m), and normally congested (vehicle spacing of 5 m), to noncongested (vehicle spacing of 60 m). The lane model distribution ratios were set according to both uniformly and nonuniformly distributed proportions of each lane model, the latter being more reflective of real-world situations. Results: The results reveal several key insights: (1) DVTDLs significantly affect evacuation time under extremely congested and normally congested conditions, with the mean difference of evacuation time reaching 43.51 s at the maximum. (2) Under noncongested conditions, DVTDLs have little effect on the average evacuation time. However, a uniform distribution of vehicle models across each lane increases the variability of evacuation times. (3) Large vehicles are likely to become stranded at the pedestrian crossing entrances, particularly in congested conditions and when vehicle types are uniformly distributed across lanes. This is especially true for large vehicles in overtaking lanes, leading to potential congestion during evacuation and, subsequently, longer evacuation times. (4) Given the same number of vehicles and vehicle type proportions, different vehicle distributions can cause significant differences in evacuation time. The maximum difference observed between the longest and shortest evacuation times was 59.76 %. Conclusions: By applying RAVT and Pathfinder software and analyzing the results, we have gained valuable insights into the effects of DVTDLs and random vehicle distribution on evacuation simulation results. We recommend that researchers and engineers consider these factors when conducting tunnel evacuation simulations. The algorithm and system developed in this study, along with the simulation method employed, provide a more scientifically robust reference for tunnel evacuation simulations.

Objective: The monitoring value of surrounding rock displacement has the characteristics of complexity and nonlinear dynamic change, and the static one-time learning of previous optimization algorithms combined with a single regression model cannot be practically applied in real scenarios. The regression fitting model uses several displacement monitoring point data to construct a general model of the surrounding rock displacement change, which cannot be applied to predict the future changes in monitoring points. The autocorrelation of the surrounding rock displacement data makes it more practical as a time series prediction problem. However, the generalization performance of a single model is easily disrupted by historical monitoring data, resulting in inaccurate prediction of test applications. In this study, a dynamic prediction method for surrounding rock displacement time series combined with time series monitoring data preprocessing is proposed. Methods: First, the displacement monitoring data of the tunnel-surrounding rock are preprocessed. The intercepted stability monitoring data are isometrized by cubic spline interpolation, and the monitoring data are decomposed into trend and random term displacement components by variational mode decomposition signal processing. Adaboost integrates 10 long short-term memory networks to construct an integrated optimization model for time series prediction. Then, the weights of the training samples are initialized, the weight coefficients of the base model in the integration are calculated by training the first base model, and the weights of the training samples of the next base model are updated. Finally, the weight coefficients of all base models are obtained. After Adaboost integration optimization, the prediction results are calculated using all base models and their weight coefficients. After training and learning, single-step dynamic prediction is performed, and monitoring changes are updated in real time to model learning. The cumulative displacement prediction results can be obtained by superimposing the trend and random term displacement sequences using the time series decomposition principle. Results: The displacement components of the rock surrounding the Central Yunnan Water Diversion Project were predicted and superimposed, and three displacement data were obtained. Compared with the traditional time series prediction model, each displacement index exhibited good performance. The complete data of the surrounding rock displacement time series were obtained by the FLAC 3D numerical simulation engineering section, and the application performance of the integrated optimization model was verified. Results showed that the integrated optimization model exhibited good performance in each component and cumulative displacement and was less affected by deformation rate fluctuation than the traditional model. Conclusions: After preprocessing the time series data, the influencing factors of surrounding rock displacement and deformation are decomposed, and multiple time series prediction models are integrated for single-step dynamic prediction, which improves the shortcomings of previous studies. The correction determination coefficient and symmetrical average absolute percentage error are used as performance indicators to verify that the prediction accuracy achieves the expected goal and is superior to the traditional classical model in solving the time series problem, which promotes the predictability of surrounding rock displacement in practical applications.

Objective: Laser profilers are widely utilized in various fields owing to their high precision, noncontact, and low cost. However, the lens plane for traditional laser profilers is parallel to the imaging plane. Thus, the high-precision measurement range of a traditional laser profiler is limited by the camera's restricted depth of view. To address this issue, this study optimizes the traditional laser profiler design and proposes a calibration method. Methods: Specifically, this study establishes a constant focus optical path in the laser profiler by tilting the lens to meet the Scheimpflug condition, wherein the imaging, lens, and light planes intersect in a single line, called the Scheimpflug line. Furthermore, the traditional imaging model is not suitable for the detection principle of the laser profiler; hence, the corresponding calibration ideas must be improved and optimized. This study proposes a complete and effective calibration method for the laser profiler, which can be divided into two parts: camera calibration and light plane calibration. For the camera calibration part, a tilt camera imaging model is established based on the traditional camera imaging model using a two-dimensional tilt angle. A method of obtaining the initial parameters and a nonlinear optimization process for the parameters are presented to rapidly obtain the tilt camera imaging model parameters. For the light plane calibration part, a calibration target, which has a double-step shape, is designed. Precise subpixel coordinates of the feature points on the laser profiler are obtained through image processing algorithms by collecting the contour image of the calibration target once the laser profiler is used. The light plane parameters are acquired using the subpixel coordinates for the least squares fitting, which quickly completes the light plane calibration. This study also designs a three-degree-of-freedom automatic calibration device to address various issues, including the removal of the laser profiler's filter, the manual adjustment of the laser profiler's pose, and the complex operating procedures in traditional calibration experiments. Results: This study used the automatic calibration device to complete the calibration and accuracy evaluation experiments and verify the correctness and effectiveness of the proposed scheme. The experimental results revealed of the following: (1) The laser profiler designed herein could clearly capture all the feature points on the light plane, thereby effectively solving the limited measurement range problem of the traditional laser profiler. (2) The reprojection errors of the laser profiler's camera were 0.487 with the traditional camera calibration method and 0.129 with the camera calibration method. (3) The calibration target could complete the light plane calibration by collecting only one image according to the expected goal. (4) After completing all the calibration steps, the average detection deviation of the laser profiler for measuring the size of the standard ceramic gauge block was approximately 0.028 0 mm. Conclusions: Thus, this study significantly improves the profiler's high-precision measurement range by establishing a constant focus optical path in the laser profiler. A calibration method with higher accuracy and efficiency is proposed herein for the laser profiler. The detection accuracy of the calibrated laser profiler meets the actual industrial requirements.

Objective: Cable cranes are the main concrete transportation equipment used in arch dam construction. Productivity analysis of cable crane transportation is crucial for improving scheduling management, reducing operational costs, and controlling dam construction progress. However, the traditional manual recording method for analyzing cable crane productivity is time-consuming and labor-intensive. Moreover, existing monitoring methods, such as sensors and global navigation satellite systems, are susceptible to interference because of the challenging environment and complicated operating space at dam construction sites. Furthermore, they usually entail high installation and maintenance costs. Therefore, this study proposes an intelligent monitoring method based on visual tracking and pattern recognition for cable crane transportation in dam construction. Methods: The proposed method initially tracks the process of cable crane transporting concrete using visual tracking technology to obtain the complete moving trajectory of crane buckets. Subsequently, it establishes a pattern recognition model to automatically identify the working states of cable cranes and calculate their productivity by analyzing the time-series features of the trajectory data. In the visual tracking of cable cranes, the main challenge is to address the similar appearance and occlusion problems of crane buckets. Therefore, we propose a new multiobject tracking framework by introducing a rematching mechanism based on tracklet features (segments of the entire object trajectory), which effectively reduces the occurrences of ID switches and enhances tracking accuracy. Additionally, You Only Look Once (YOLO) model is trained as the object detector of the tracking framework. Subsequently, trajectory data obtained by visual tracking is used as input for the pattern recognition model of cable crane working states, with the output being the pouring productivity. This pattern recognition model employs spline interpolation and Savitzky-Golay filters to solve the problems of missing values and noises in the trajectory data. A first-differential method is applied to statistically analyze the variation patterns of the trajectory data. This model can rapidly and accurately identify the working states and determine the key efficiency indicators of cable cranes. Results: A testing experiment was conducted at an arch dam construction site to evaluate the monitoring performance using this approach. Experimental results are summarized as follows: 1) The proposed vision-based multiobject tracking method proves effective in detecting and tracking cable buckets in intricate construction scenes, thus achieving effective and complete tracking of moving trajectories of crane buckets; moreover, identity F1 score (IDF1) and multiple object tracking accuracy (MOTA) metrics reach 94.8% and 90.0%, respectively. 2) The proposed pattern recognition model can rapidly and accurately distinguish six working states in the cable crane transportation process, including horizontal transport, descent, unloading, ascent, horizontal return, and waiting for loading. 3) Key productivity indicators, such as duration of a single transporting cycle, number of transporting cycles, duration of each working state, and concrete pouring intensity, are accurately calculated and meet engineering management requirements. This also confirms the practicability, reliability, and accuracy of the proposed monitoring method. Conclusions: Thus, this study successfully integrates vision-based tracking and pattern recognition technologies to develop an intelligent monitoring method, consequently achieving automatic and accurate calculation of cable crane productivity. Furthermore, it demonstrates a positive application effect at dam construction sites and provides innovative perspectives and technical support for construction management.

Objective: Discontinuities are vital components of rock mass, significantly affecting its strength, deformation, and seepage characteristics. They provide essential parameters for rock mass classification and engineering design. Borehole television technology is a widely used method for capturing these discontinuities within the rock mass, offering high-resolution in situ images. However, in complex strata, the discontinuities appear in various morphologies with significant width differences. Coupled with the rapid texture changes on rock wall faces, these discontinuities create a highly uneven contrast, making it challenging for traditional algorithms to recognize them accurately. To address this challenge, this study introduces an improved deep learning network model specifically designed for borehole images of complex strata. Methods: The proposed model, based on the U-Net architecture, incorporates a deeper encoding-decoding network structure. This structure effectively handles semantic information related to discontinuity breaks caused by uneven contrast. The model integrates channel attention mechanisms and residual modules, enhancing feature extraction capabilities at different levels in the encoding stage. In addition, the channel attention mechanism fuses multichannel discontinuity information from both encoding and decoding layers. A multiscale spatial attention module introduced in the lower-level skip connection improves the ability to process complex morphological discontinuities and enriches the semantic features of discontinuities in the coding layer. In this study, the borehole image data are augmented in various ways, such as using perspective deformation similar to the stratum deformation under geological action. This study also employs joint training with focal loss and Dice loss to handle imbalanced image data. The generalization ability of the network model is thoroughly validated through ablation studies and comparative experiments using the same borehole image as the training set and neighboring borehole images as the test set. For comprehensive quantitative evaluation, this study uses several metrics, including precision, recall, F_1-Score, and F_2-Score. Results: Our experimental evaluation, conducted on a self-made borehole image dataset, indicated that compared to several common image segmentation network models, our proposed model significantly improved the recognition capability of rock discontinuities in borehole images from complex strata while ensuring faster computational efficiency. The precision and recall on the test set for the proposed model reached 78.23% and 77.85%, respectively. This marked an improvement in segmentation performance by 7.96% and 14.99%, respectively, compared with the basic U-Net model. Both the F_1-Score and F_2-Score were close to 78%. Although the model size was 18.13 MB and had approximately twice the parameters of the base U-Net, the deeper network hierarchy reduced the number of channels of shallow high-resolution feature maps, resulting in a reduction in computational load. The model achieved an FPS of 85, which was slightly higher than that of the basic U-Net model. Conclusions: This study meticulously improves upon the basic U-Net model by strategically incorporating the attention mechanism, residual connections, and multiscale convolutions. The improved model exhibits high accuracy and robustness. It effectively confronts the challenges associated with balancing detailed features and high-level semantics owing to significant width differences in discontinuities within complex strata. Furthermore, it addresses issues related to incomplete extraction of discontinuities caused by uneven contrast between discontinuities and rock wall surfaces. As such, this improved model provides strong technical support for the automatic identification of rock discontinuities in on-site borehole investigations.

Significance: The construction industry in China is a major contributor to carbon emissions, creating substantial environmental challenges. In response, the construction sector is intensifying efforts to reduce its carbon footprint. Among the various strategies implemented, building information modeling (BIM) technology has emerged as a key digital tool with transformative potential to lower building-related carbon emissions. BIM technology enhances design precision and operational efficiency while enabling comprehensive analysis and optimization of building systems. This capability facilitates carbon emission reductions throughout the lifecycle of a building. However, there remains a notable lack of systematic documentation and synthesis on effectively leveraging BIM technology for carbon emission control in construction. This gap is further exacerbated by the lack of comprehensive analyses of potential future research directions and practical application scenarios for BIM in carbon reduction. Progress: Therefore, the present study investigates the specific application of BIM to reduce carbon emissions across the design, production, and operation phases of a building's lifecycle. Through bibliometric methods that entail quantitative analysis of published research, the study seeks to identify key technologies and emerging trends within this domain. This research is organized into two main components. First, a comparative literature review combined with a market survey is conducted to map advancements in BIM-based research related to the whole life cycle carbon emissions of buildings. This comprehensive review aims to consolidate existing knowledge while identifying gaps or inconsistencies within the current body of research. Second, a detailed examination is conducted, focusing on the stages that have the most significant impact on carbon emissions, including building design, production, and operation. This analysis aims to identify major achievements and ongoing challenges within current research efforts and practical implementations and highlight potential directions for future advancements. Conclusions and Prospects: The findings reveal several key insights. BIM technology has focused primarily on the whole life cycle carbon emission analysis and design phase of buildings. While these contributions are noteworthy, research targeting the production and operational phases remains comparatively underdeveloped. This imbalance is partly due to the limited exploration of BIM's application scenarios in these later stages of a building's lifecycle. Specifically, BIM's potential to optimize building production processes and enhance operational efficiency through real-time data analytics and predictive modeling has not been completely realized or integrated into practical projects. Therefore, future research should prioritize broadening BIM's application to cover all phases of a building's lifecycle comprehensively. This involves developing innovative BIM tools and methodologies that seamlessly integrate with building management systems to enable real-time monitoring and control of carbon emissions. Furthermore, fostering collaboration among academia, industry stakeholders, and policymakers is essential for advancing BIM-based carbon reduction strategies and ensuring their effective implementation in practical scenarios. By addressing these research and implementation gaps, the construction industry can fully leverage BIM technology to achieve substantial reductions in carbon emissions, thereby contributing to global sustainability efforts.

Objective: Named entity recognition (NER), a central task in the information extraction realm, aims to precisely identify various named entity types in textual content, including personal names, locations, and organizational names. In Chinese NER domain, deep learning techniques are crucial for character and vocabulary representations and feature extractions, yielding remarkable research achievements. Common deep learning models for NER include sequence labeling, span-based approaches, generative methods, and table-based strategies. Nevertheless, this task suffers from the scarcity of lexical information. Hence, this challenge is perceived as a primary hindrance limiting the development of high-performance Chinese NER systems. Despite developing extensive lexical dictionaries encompassing rich vocabulary boundaries and semantic insights, effective incorporation of this lexical knowledge into Chinese NER task remains a considerable challenge. Particularly, the seamless integration of semantic information from matching vocabulary and its contextual cues into Chinese character sequence remains intricate. Moreover, ensuring the accurate delimitation of named entity boundaries is still a remarkable concern. Methods: In the realm of intelligent judicial systems, the NER task within legal documents has garnered significant attention. Nonetheless, prevailing sequence labeling models predominantly rely on character information, constraining their capacity to capture semantic and lexical contextual nuances and inadequately addressing entity boundary constraints. To resolve these challenges, this paper introduces an innovative model called semantic and boundary enhanced named entity recognition (SBENER). To enhance the semantic features of legal documents within the SBENER model, external information containing vocabulary pertinent to theft crimes is smartly integrated. Initially, word vectors for theft crime terms are acquired through pretraining. Subsequently, a vocabulary dictionary tree is constructed, enabling the potential vocabulary candidate identification for each character. Further, these candidates are amalgamated into a final external information vector via a bilinear attention mechanism. Additionally, a linear gating structure is introduced to mitigate interference from external information in the original text. To overcome the limitations of sequence labeling models for managing entity boundary constraints, this study designs a boundary pointer network within the model to confine entity boundaries. This involves embedding the character sequence into hidden layer representations via bidirectional long short-term memory followed by decoding to introduce probability constraints for each entity span. Ultimately, contextual and boundary information is inputted into a conditional random field for obtaining the ultimate entity classification outcomes. This design adroitly tackles the issues of vocabulary loss and boundary constraint scarcity within sequence labeling models. Results: Experimental results on the CAILIE 1.0 and LegalCorpus datasets corroborated the effectiveness of the proposed method, yielding F₁ scores of 88.70% and 87.67%, respectively, surpassing other baseline models. Additionally, the study conducted ablation experiments to validate the effectiveness of each component. The experimental results showed that integrating external semantic information related to theft, enhancing entity boundary constraints through pointer networks, and incorporating gating mechanisms to restrict irrelevant information fusion were all effective approaches for achieving higher F₁ scores for the model. Furthermore, this paper applied dimensionality reduction to external semantic word vector information and conducted experimental analysis on different fusion layers. Single-layer fusion outperformed multilayer fusion, while fusion at intermediate levels yielded better results. This underscored the marked enhancement in judicial NER facilitated by the proposed approach. Conclusions: The SBENER model effectively enhances the proficiency of recognizing named entities in legal documents through the fusion of external information and reinforcement of boundary constraints. This pioneering method substantially contributes to advancements within the intelligent judicial systems.

Objective: An accurate measurement of the heat release rate (HRR) of a fire source is crucial for thoroughly understanding the fire evolution process. However, the commonly used oxygen consumption method requires expensive equipment, leading to high operational costs. According to the energy conservation principle, heat released per unit of time during material combustion is closely related to the increase of the room temperature. Machine learning methods have demonstrated considerable potential for exploring relationships between several independent and dependent variables, and attempts have been made to predict fire parameters based on temperature. Therefore, based on previous studies, this paper proposes a comprehensive machine learning framework to predict the HRR using temperature data as input. Furthermore, feature selection techniques are innovatively introduced to obtain key location temperatures to maximize HRR prediction accuracy. Methods: First, fire scenarios with different parameters are simulated in an ISO 9705 room using the fire dynamics simulator (FDS) software. Temperature data at different locations are obtained by gridding thermocouples, and a fire database is constructed. Subsequently, feature selection using recursive feature elimination (RFE) algorithms based on least absolute shrinkage and selection operator (Lasso) and random forest (RF) is performed to obtain two different low-dimensional subsets from high-dimensional simulated temperature features. Control groups with the same number of features are established. Finally, the performance of three typical models, namely, linear regression (LR), K-nearest neighbors (KNN), and light gradient boosting machine (LightGBM), for predicting the HRR are compared using different feature subsets. Results: The results show that using the subset obtained by RFE based onRF, the LightGBM model demonstrates the lowest root mean square error (RMSE) and mean absolute error (MAE) values of 23.89 kW and 15.49 kW, respectively, indicating the least difference between its predicted and observed values. Furthermore, regarding the coefficient of determination, the LightGBM model reaches the highest value of 0.991 6, close to 1, thereby demonstrating its superior fitting capability. This is primarily due to the complex nonlinear relationship between the trained temperature dataset and HRR. Compared to the KNN and LR models, the histogram algorithm and the leaf-wise growth strategy with a depth limit enable the LightGBM model to give full play to its advantages. Tree-based models, such as LightGBM and XGBoost, can also be used for algorithm model-level improvements in the future. Additionally, deep learning models, such as multilayer fully connected and convolutional neural networks, can be utilized for fitting complex mappings. Compared with LightGBM models trained with manual feature subsets, RFE based on RF decreases the prediction errors (RMSE and MAE values decrease by 46.54% and 50.66%, respectively). The coefficient of determination also increases by 2.1%, validating that this feature selection method can considerably improve prediction accuracy over manual feature selection. Conclusions: This paper proposes a comprehensive machine learning framework to obtain thermocouple temperature through the FDS fire simulation and then combines the feature selection and prediction models for HRR prediction, substantially improving prediction accuracy over manual feature selection. This comprehensive framework has theoretical and practical significance, thereby opening new pathes for HRR prediction. Subsequent research studies will construct additional comprehensive fire databases, increase the number of feature parameters, or explore algorithm combinations to improve HRR prediction.

Objective: Machine learning algorithms provide valuable data support for designing and optimizing low-carbon residential buildings. However, when used directly for carbon emission prediction and analysis, these models often lack proper parameter tuning and optimization. The different impacts of various independent variable datasets on predictive performance also remain to be clarified. In China's cold zones, where residential buildings share similar architectural structures, energy-saving designs, and spatial layouts, carbon emissions primarily come from the operational phase and the production stages of building materials, with heating emissions being a significant component. This study aims to elucidate the effectiveness of different machine learning algorithm models in guiding low-carbon residential design in these cold zones, offering architects criteria for selecting proper algorithms. This study focuses on automatic parameter tuning and optimization for several commonly used algorithms in the context of low-carbon design of buildings, including multiple linear regression, classification and regression tree, random forest, adaptive boosting, gradient boosting regression tree, and multilayer perceptron. The study compares and analyzes the performance limits and applicability of these algorithms and independent variable datasets in predicting carbon emissions during building material production and heating stages. Methods: This paper elaborates on the target boundaries, parameter ranges, optimization processes, and validation methods for optimizing machine learning algorithm models. Through comprehensive research and simulation analysis of 37 reinforced concrete shear wall residential buildings and their derivative schemes in cold zones, multiple independent variable datasets suitable for establishing predictive models are identified. Cross-validation and grid search techniques are employed to optimize the predictive performance limits of different machine learning algorithms and independent variable datasets. Subsequently, 120 models for predicting carbon emissions from building materials and 60 models for transforming steady-state heating consumption into dynamic heating consumption using the six mentioned algorithms are established. Results: A horizontal comparison of the models reveals that algorithms such as multiple linear regression, random forest, and gradient boosting regression trees exhibit relatively good performance (R² over 0.900) in carbon emission prediction after hyperparameter tuning across different independent variable datasets. Random forest and gradient boosting regression tree models excel in error control and offer similar predictive accuracy to multiple linear regression but lack interpretability. In contrast, multiple linear regression models provide clearer equations and stronger guidance for low-carbon design and optimization, focusing on carbon emission reduction during building material production or winter heating stages. Models based on the total residential building area exhibit optimal performance in predicting building material carbon emissions. Predictive models built on parameters such as the number of above-ground and underground floors, building width and depth, total household numbers, number of bedrooms for standard floor, and total number of residential bathrooms in the residence also demonstrate strong predictive capabilities for building material carbon emissions. For predicting the conversion coefficient during the heating stage, including the number of households and bedrooms per standard floor as independent variables significantly enhances predictive performance. Conclusions: Although various machine learning models are useful for predicting residential building carbon emissions, the multiple linear regression model stands out owing to its excellent predictive performance and its intuitive representation of how design parameters affect carbon emissions. By utilizing different and appropriate independent variable datasets, such as the total number of floors, floor height, building dimensions, number of households and bedrooms on a floor, and corrected coefficients for urban meteorological parameters (including outdoor average temperature during the heating season, actual heating days, and roof and wall heat transfer coefficients), or by adopting the finally determined total building area, the multiple linear regression algorithm can deliver timely and multi-faceted guidance. These results are crucial for low-carbon design and optimization during the primary stages of the residential lifecycle in China's cold zones.

Objective: Safe and effective evacuation of individuals during extreme weather conditions is critical in evacuation planning. Compared with traditional personnel evacuation and emergency transportation studies, evacuation zone planning is in its early stages. It lacks comprehensive consideration of major urban disaster scenarios, particularly beyond hurricanes. Additionally, there is no unified system for defining problems or measuring urban population hotspots, spatiotemporal disaster impacts, and exit distribution in evacuation planning. To address the practical issues of evacuating affected individuals during heavy rain and flood disasters, this paper proposes a model for delineating evacuation zones. Methods: Starting with establishing evacuation needs and quantifying the impact of disasters on road segments, this study considers urban population distribution hotspots and the characteristics of heavy rain and flood disasters. Through modeling analysis, geographic information system (GIS) visualization, and other methods, a model is developed for the integrated delineation of evacuation zones and the allocation of evacuees at exits. The main components include the following: (1) To identify hotspot areas in urban functional zones and establish evacuation needs based on the city's road network. (2) To assess the risk of heavy rain and flood disasters, establish a risk indicator system for flood risk assessment (including causative factors, disaster-prone environments, disaster-prone bodies, and disaster prevention and mitigation capabilities), and develop a road damage model to determine road network damage. (3) To construct a two-tier planning optimization model to determine evacuation paths and exit allocations. (4) To use Wuhan's Wuchang district as an example, the effectiveness of the proposed method for large-scale urban evacuation zone planning under flood disasters is validated. The upper-level model provides the proportion of evacuees that each evacuation point should accommodate, with these allocation ratios stored in chromosomes as input for the lower level. The lower-level problem uses the incoming allocation ratios to calculate the evacuation flow for each OD pair and evaluates the fitness of the upper-level chromosomes. This is achieved using the Frank-Wolfe algorithm. The two-tier framework allows for detailed treatment of complex evacuation planning problems, ensuring the global minimization of total evacuation time and individual minimization of evacuee travel time. Results: The innovative aspects included identifying evacuation needs in urban hotspots and constructing road damage levels under risk zoning for heavy rain and flood disasters. The two-tier planning optimization model minimized overall evacuation time and individual travel time, making the evacuation plan more realistic and reasonable. Conclusions: The proposed method for large-scale urban evacuation zone planning is feasible, risk assessment is essential in actual evacuation planning. Significant differences exist in day and night population distribution with daytime populations primarily concentrated in commercial and work areas and nighttime populations concentrated in residential areas. Emergency management departments should develop varied evacuation plans for different periods. Due to potential road damage during disasters, preplan alternative evacuation routes and make real-time dynamic adjustments during evacuations.

Objective: The Yarlung Zangbo River contains numerous hydropower resources, with high head and large flow rates in the downstream region, which is conducive to power generation by Pelton turbines. Pelton turbines convert the kinetic energy generated from the water potential energy into mechanical energy for rotating a runner. The runner is the core component for the flow and work of the Pelton turbine, and the shape of its bucket is crucial for the runner's performance, which is uniformly arranged along the hub of the runner. As the surface shape of the bucket is complex, several parameters are required to determine its geometry model, undoubtedly posing a huge obstacle to work. Methods: In this paper, a design method is proposed to address the problem of designing and improving the bucket based on the Bézier curve. The design space is simplified as much as possible based on geometry, and the Bézier curves are utilized for designing the bucket shape. An orthogonal analysis is applied for the optimization of bucket parameters, while the computational fluid dynamics method is employed for analyzing the energy characteristics and three-dimensional flow field of the Pelton turbine. In the bucket design method, the three-dimensional geometry of the bucket can be divided into contour, flow profile, and guidelines, and several characteristic parameters can be determined for those lines. Each type of line includes several biquadratic Bézier curve connections. The number of characteristic line parameters is decreased by establishing a connection between five control points of the Bézier curves. Thus, a three-dimensional design method for the bucket of the Pelton turbine is proposed based on the five controlled characteristic parameters. The main optimization parameters are chosen by the geometry. Subsequently, bucket depth, width increment, outflow angle, splitter angle, and cutout diameter are chosen to conduct orthogonal optimization for the Pelton turbine bucket. For further analysis of the flow characteristics of the optimized bucket, the runner is modeled based on the optimum parameters. In the computational fluid dynamics method, grids are meshed by ICEM, and computational fluid dynamics is performed with ANSYS FLUENT. Results: The results of the polar analysis and three-dimensional unsteady flow field revealed that width had the maximum influence on runner efficiency; outflow angle, cutout diameter, and bucket depth had a smaller influence; and splitting angle had the minimum influence. After optimization, the hydraulic efficiency of the Pelton turbine was increased by 6.71%. The optimized bucket demonstrated a larger torque peak than the prototype bucket. The bucket always showed large torque when its torque decreased to zero and exhibited smoother curve transition and longer work time. Thus, the optimized bucket demonstrated greater total torque than the prototype bucket; furthermore, the former's high-pressure area was larger, making the energy conversion of water from the nozzle to the bucket more effective. Conclusions: This paper proposes a three-dimensional design method for the Pelton turbine bucket based on the controlled characteristic parameters. The energy performance of the Pelton turbine was enhanced by the orthogonal optimization and three-dimensional flow simulation.

Objective: Most accidents in the construction industry are caused by hazards that remain unrecognized due to inexperience and inattentiveness. Novice workers have difficulty learning how to quickly and accurately determine the source of a hazard and avoid it; thus, it is necessary to develop a dynamic hazard identification process-assisted education pattern. Eye-movement modeling examples (EMMEs) are videos of gaze replays of hazard identification trajectory by an expert with a verbal explanation. Methods: This study constructed the EMMEs of hazard identification by an expert to explore the mechanism of its influence on workers' safety education at different experience levels. We created eight virtual construction sites for hazard identification testing, which mainly included falls, collapses, electric shocks, lacerations, explosions, and unsafe actions. A participant's task was to search for hazards, i.e., to visually inspect construction site scenarios and determine where a safety accident might occur. An eye tracker was used to collect the search patterns of experienced and novice workers before and after EMME training. Eye movement data were collected from 14 novice workers and 10 experienced workers. The study followed a 2×2 mixed-group design with between-subject factor experience (experienced vs. novice workers) and a within-subject factor case (before vs. after EMME training). Hazard identification accuracy, task completion time, and sequence standardization were used as indicators to measure the identification performance of the participants before and after EMME intervention. Results: Herein, a t-test was used to evaluate the difference between the hazard identification performances of novice and experienced workers, and the interaction effect was used to test the moderating effect of EMMEs on prior experience and hazard identification performance. The main results were as follows: (1) Participants with EMME intervention performed better at hazard identification and showed higher hazard identification accuracy, shorter task completion time, and higher sequence standardization after EMME training. This finding confirmed that instructions comprising EMMEs effectively improved construction safety education. (2) The hazard identification performance of experienced workers was better than that of novice workers in the pretest; compared to novice workers, experienced workers identified more hazards in less time with more standard sequences before EMME training. The experienced workers consistently inspected laborers first, then the equipment or environment, and finally, the entrance. Novice workers typically inspected the hazards in the same order but with a less consistent scan path. (3) The EMME-based safety education mode had the expertise reversal effect. Participants with rich work experience showed insignificant improvement in performance after EMME training, while novice workers benefited far more from EMME intervention than experienced workers. Conclusions: Our results demonstrate the potential of EMMEs to indirectly teach strategic hazard identification sequences and contribute to deeper safety education, particularly for workers with limited work experience. This study has educational importance as it provides new evidence of the potential of eye-tracking technology as an indirect instruction tool.