The legal field is using more artificial intelligence methods such as legal judgment prediction (LJP) based on case description texts using natural language processing. Charge prediction and law article recommendations are two important LJP sub-tasks that are closely related and interact with each other. However, previous studies have usually analyzed them as two independent tasks that are analyzed separately. Furthermore, charge prediction and law article recommendations both face the problem of confusing charges. To this end, this paper presents a multi-task learning model for joint modeling of charge prediction and law article recommendations. Confusing charges are handled by using a set of charge keywords extracted from case description texts using statistical techniques for integration into the multi-task learning model. This method was evaluated using the CAIL2018 legal dataset. The results show that incorporating the charge keywords into the multi-task learning model effectively resolves the confusing charge problem and significantly improves both the charge prediction and the law article recommendation results.

This article presents a neural network model for crime prediction, legal article recommendation, and sentence prediction from judicial documents. The model is based on a hybrid attention and CNN model which combines the improved hierarchical attention network (iHAN) and the deep pyramid convolutional neural network (DPCNN) by ResNet. The F1-Scores (mean value of Micro-F1 and Macro-F1) for the crime prediction and related law samples from CAIL-2018 were 85% and 87%. The sentence prediction accuracy is impacted by differences in locations, dates, courts, judges, and defendant attitudes. The model adjusts well to these differences because of its high predictive ability and model generalization. The model prediction outputs for the recommended crime prediction and law items were then added to the model input for the sentence prediction task to further improve the model performance. The model got an excellent result in the sentence prediction task (CAIL-2018) with an F1-Score of over 77%.

Integrated terrestrial-satellite networks (ITSN) are a major project of China's "Scientific and Technological Innovation 2030" and an important development direction for the Internet. The inter-domain routing protocol is a key factor for effective integration of various satellite networks and existing terrestrial networks in ITSN. However, the network topology of satellite network changes frequently, which presents significant challenges for deployment of the existing inter-domain routing protocol. The widely used inter-domain routing protocol border gateway protocol (BGP) is used here to develop a network-topology-decoupled BGP (NTD-BGP). The protocol uses an inter-domain neighbor discovery scheme that is independent of the physical topology changes and a module for active routing updates. The protocol uses the predictability of the satellite movements to decouple the mapping between the network topology and the inter-domain neighbor relationship as well as routing updates to achieve efficient, stable inter-domain routing in ITSN. Emulations show that this protocol more effectively maintains inter-domain neighbor relationships, more effectively reduces routing update times, and more effectively maintains network stability than the traditional BGP.

Deep learning based network anomaly detection is a new research field with previous studies using preprocessed datasets based on data mining or other methods. This paper transforms and encodes the UNSW-NB15 dataset using one-hot encoding to a two-dimensional dataset. Then, GoogLeNet is used for deep learning network to extract the features and train the classifier. Tests show that this method can effectively process the original network packet with a classification accuracy over 99%, which is much higher than deep learning detection methods based on preprocessed data.

Synthesized speech quality can be significantly improved by correctly dividing the prosodic structure of sentences. The feature selection is then one of the key factors of prosodic structure prediction. In Chinese information processing, the text features can be divided into shallow text features and deep text features, with the shallow features including words, parts of speech, word length and other factors while the deep features include syntactic information, semantic information and other factors. The relationships between the syntactic dependency structure and the prosodic structure were analyzed to identify the shallow and deep text features in the text with a conditional random field model used for prosodic phrase prediction. This study first uses the shallow text features to recognize the prosodic phrases and then adds the syntactic dependency deep text features to construct the model. Tests show that the accuracy is increased by 13.3%, the recall rate is increased by 14.69%, and the F-score is increased by 14.1%.

The traction force of wheeled horizontal-well tractors is mainly produced by the contact between the wheels and the pipe walls. Previous studies ignored the plastic deformation caused by the contact between the wheels and the walls and used empirical friction coefficients to model the forces, which has led to significant errors. This paper derives a wedge indentation model (the wedge symmetry axis is perpendicular to the pipe wall) based on the simultaneous action of the normal force and the torque to predict the indentation formation according to slip line theory. Finite element analysis results not only show the plastic deformation for various forces, but also verify the quantitative relationship between the forces and the indentation shape. Comparisons with measured indentation shapes shows that the theoretical model can predict the indentation depth and the pile-up height, which can be used to improve sideling indentation modeling of wheels for better traction modeling.

Lower limb exoskeletons (LLEs) can assist paraplegic or hemiplegic patients to regain movement. However, different physiological structures and assembly errors create undesirable forces between the human and the exoskeleton, which result in undesired loadings on the joints and soft tissues of the user. Traditional physical human-robot interfaces of LLEs are unable to solve this problem. This study developed a physical human-robot interface for LLEs that focused on the internal joint forces through comparisons with traditional physical human-robot interfaces. A statics analysis of the forces on the human lower limb during the swing phase showed that this device stabilizes the internal joint forces. The device reduces the internal joint force on the human hip by 2.45% on average when there is vertical difference in the sagittal plane between the human hip joint and the exoskeleton hip joint and by 11.23% on average when there is horizontal difference in the sagittal plane between the human hip joint and the exoskeleton hip joint. The device also reduces the force on the human knee by 13.68% when there is vertical difference in the sagittal plane between the human knee joint and the exoskeleton knee joint, but increases the force on the human knee by 23.48% when there is horizontal difference in the sagittal plane between the human knee joint and the exoskeleton knee joint. This paper shows that this device can reduce the wear on the human hip and knee cartilage and soft tissues; thereby improving the exoskeleton experience for patients.

This study investigated the intellectual property concerns related to recycling spent lithium-ion batteries in China. This report summarizes the technical development trends, main regional distribution networks, critical technical fields, core technical points, and the core applicants. The results show that in the past 3 years, the number of patents related to lithium-ion battery recycling has increased rapidly to over 200 patents/a. There are many cross-disciplinary patents with many patents focused on the electrical components, metallurgy, chemistry, and environmental effects. The main applicants are not just limited to universities and research institutes, but also include a large number of companies since this research has significant industrial value. The stronger national environmental requirements and the increasing demand for resources is leading to many improvements in recycling methods as indicated by the increasing number of patents. Industrial policies, especially pollution prevention and control standards, provide important incentives for further research and for the development of improved recycling methods.

The effects of different pretreatments including ultrasound, microwave and alkali thermal pretreatments on anaerobic co-digestion of kitchen waste and other organic wastes were studied through biochemical methane potential (BMP) experiments. The experiment without pretreatment served as the control. After the pretreatments, some of the primary organics are solubilized by hydrolysis and liquefaction reactions resulting in higher concentrations of volatile fatty acids (VFA), soluble chemical oxygen demand (SCOD) and total ammonia nitrogen (TAN) which improve the efficiency of the subsequent anaerobic co-digestion. The BMP results show that the pH and VFA, SCOD and TAN concentrations in the four experiments all have the same trend and gradually stabilize after 6 days, indicating that the biodegradable organics in the substrate are completely decomposed in a short time. In addition, the pretreatments improve the biogas production rate. Ultrasonic pretreatment increases the methane yield from 285 mL/g VS to 324 mL/g VS, which is consistent with the modified theoretical methane yield, while microwave and alkali thermal pretreatments inhibit anaerobic fermentation.

A control algorithm verification platform was built based on a rapid prototype controller to provide fast, effective verification of magnetorheological semi-active suspension control algorithms. The kingpin angle and the space angle of the lower arm were considered in the 1/4 vehicle suspension system model. The acceleration sensor and the body height sensor were mounted on the suspension. An LMS Test Lab system was used to collect and analyze the suspension system vibration data. The rapid prototype controller was used as the carrier for developing and debugging the control algorithms. Platform verification tests were used to verify the algorithm development, the control process monitoring, the control algorithm adjustments, and the quantitative analyses. This platform meets the design goals and is convenient, fast, and effective for developing, debugging, and verifying semi-active control algorithms.

When a commercial truck turns while stationary, the power steering system needs to ensure that the driver can turn the wheels at this moment with the greatest steering resistance torque. Therefore, the stationary steering resistance torque of a truck is an important factor when designing power steering systems. The LuGre friction theory model for the friction resistance between the tire and the ground was used with models of the whole vehicle and its steering system to predict the gravity aligning torque and the steering friction resistance torque. The predictions gave the stationary steering resistance torque and the relationship between the torque and the steering wheel angle of commercial trucks. The results agree well with experimental data showing that this model accurately reflects the relationship between the stationary steering resistance torque and the steering wheel angle for improved power steering system designs.

The multiple-relation-time (MRT) lattice Boltzmann method with a high-density-ratio two-phase model was used to simulate liquid water transport in the gas diffusion layer (GDL) and gas channels of a high-current-density fuel cell. The results show the effects of Reynolds number, perforation shapes and locations in the GDL and the angles of the wave-like gas channels on the water transport. The results show that the GDL and the gas channels should be optimized together to improve the water removal rate. In addition, the results show that the water begins running out of the GDL at earlier times as the Reynolds number increases with the times not related to the wave-like gas channel angle or the perforation shape or location. The structural optimization of the perforated GDL and the wave-like gas channels can guide future designs of fuel cells with high current densities.

The kinetic energy stored in a flywheel depends on its rotational inertia and linear velocity. The use of carbon fiber composite materials can effectively improve the rotor speed and increasing the rotor length can increase its rotational inertia. Furthermore, multi-section flywheel rotors formed by connecting tubes through bellows enables the rotor to pass through critical frequencies at low speeds, which increases the working speed range. This paper studies the bending stiffnesses of variable thickness carbon fiber bellows with the static bending stiffness measured by the axial force method. A finite element model was then used to predict the linear and geometric nonlinear bending stiffness for various loads and various laminates. The research indicates that bending stiffness variation is nonlinear with the load, with the measured stiffness being of the same order of magnitude as the theoretical values for large load. The simulations also show how the thickness, angle and stacking sequence of the laminate affect the bending stiffness.