A fault diagnosis method based on a kernel extreme learning machine (KELM) was developed to analyze thruster failures in hypersonic aircraft reaction control systems (RCS). The parameters and kernel function were optimized for faults involving aircraft actuator failures. Results using this fast, accurate diagnostic method show that the method is not dependent on the aircraft model and provides fast and accurate diagnoses of aircraft actuator faults using a data-driven process.

Data race is a typical concurrency bug in multithreaded programs. Data race is difficult to detect due to the uncertain interleaving in multithreaded programs. A random forest instruction level data race detection model is developed for multithread programs using five attributes to identify the data race features. Firstly, data race detection at the instruction level is based on the happens-before relationship and the lockset algorithm. At the same time, the assembly source code is used to eliminate implicit synchronization pairs. Then, the analysis results from the happens-before relationship and the lockset algorithm are used to train a random forest detection model for multithreaded program data race detection. This data race detection tool for multithreaded programs, AIRaceTest, is implemented on Pin. The model is trained with the results of the multithreaded program instrumentation in GitHub as a sample set. The model accuracy reaches 92.1%. Test results on the classic multithreaded programs, Google data-race-test and Parsec benchmark 3.1, show that the false positives are reduced by about 10.6% and the false negatives are reduced by about 12.3% compared with Eraser, Djit+and Thread Sanitizer. For a large number of threads, the time overhead is reduced by 41.8% while the memory overhead is reduced by 22.4%.

Human activity recognition based on wearable sensors has been widely used in various fields, but complex human activity recognition based on multiple wearable sensors still has many problems. These problems include the incompatibility of many signals from multiple sensors and the low classification accuracy of complex activities. This paper presents a multi-sensor decision-level data fusion model using multi-task deep learning for complex activity recognition. The model uses deep learning to automatically extract the features of the original sensor data. In addition, the concurrent complex activities are divided into multiple sub-tasks using a multi-task learning method. Each sub-task shares the network structure and promotes mutual learning, which improves the generalization performance of the model. Tests show that the model can achieve a 94.6% recognition accuracy rate for cyclical activities, 93.4% for non-cyclical activities, and 92.8% for concurrent complex activities. The recognition accuracy rate is on average 8% higher than those of three baseline models.

The increasing demand for real-time data stream processing is driving the development of distributed stream processing systems. The large amount of skew data streams and the heterogeneity of complex distributed systems pose challenges to the current grouping strategies of distributed stream processing systems. The existing distributed stream processing grouping strategies usually focus on balancing the number of tuples between parallel instances, while ignoring the impact of system heterogeneity on the grouping strategy. This paper presents a time-aware grouping algorithm that analyzes the network heterogeneity and the processing capability in a distributed stream processing system that considers the processing time of each downstream operator instance in the stream processing system. The algorithm also takes into account the communication time between the upstream and downstream operators with various routing strategies formulated according to the frequency of the key, so that the system achieves load balancing with little overhead. Tests on an Apache Flink distributed stream processing system show that the time-aware grouping algorithm increases the throughput by 10% while the average processing latency is reduced by 33% compared to the existing grouping algorithm.

Bug reports are used to identify and track defects for improving software quality. Software testing often uses multiple users and parallel testing. The resulting numerous bug reports must then be integrated while removing fake or duplicate bug reports. This paper presents an automatic detection method for bug reports based on the BM25 algorithm. After preprocessing the bug reports, a matching library is built based on the test requirements and test report samples. The BM25 algorithm is used to calculate the similarities between reports to identify accurate bug reports. Tests with software test contest data show that the model can correctly judge most bug reports to effectively improve the efficiency of identifying false negatives and duplicates.

Integrated circuit testing produces huge quantities of test data in expensive tests. This paper describes use of the minimum run-changing point mark coding compression method to compress the original test data to reduce the test costs. The method encodes the test set into several vectors and uses the overlapping relationships between the vector run range in each group to merge the run-length switching points. The run position in all the test vectors in the group can be represented by a vector which simplifies traditional coding compression. The method breaks through the limitation of using the suffix of the code word to express the length of the run, and greatly shortens the code word compared to conventional code compression. The method has simple decompression rules and low hardware overhead. Tests with the ISCAS 89 standard circuit experiment show that this compression scheme is more effective than other types of coding compression schemes.

The impacts of underlying urban surfaces on rainfall are due to the specific dynamics and thermodynamics properties of urban areas. The weather research and forecasting model (WRF) was used in this study to simulate the heavy rainfall event on July 20, 2016 in the Xiongan New Area as an example. Different urban canopy roughness scenarios were created to study the effects of the urban canopy roughness on the storm evolution and rainfall area. The results show that the urban canopy roughness significantly impacts the storm evolution and rainfall area. When the urban canopy is not very rough, the air is heated as it passes through the urban area, resulting in water vapor convergence and rain island effects which increase the rainfall over the urban area and the downwind region. When the urban canopy is rough, the storm elements move slowly with more rainfall upwind. Increases in the urban canopy roughness may change the direction of the storm movement and increase the rainfall outside the city. This study assesses the impacts of urban canopy roughness on rainfall and provides a scientific basis for future urban planning and construction. In addition, the simulation results have reference significance for the selection of urban rainfall simulation parameters.

Reverse flow routing is widely used in hydraulic engineering calculations. However, numerical attenuation along the flow direction results in continuously amplified errors when directly solving the hydrodynamic equations in reverse which can lead to unstable calculations. This study contrarotates the x-t plane by 90° from the forward flow position and then derives the reverse flow routing model. On the rotated x-t plane, the initial and boundary conditions for the forward routing model are transformed to the boundary and initial conditions for the reverse model, which avoids the instabilities and non-convergence caused by reverse iterations in time and space. After validation, the reverse model is applied to a real-world case in the Xijiang River. The results show that this model is accurate and stable and can effectively reconstruct the historical inflow hydrographs for various flow cases. This research can improve hydraulic engineering studies using reverse flow routing.

Stereoscopic particle image velocimetry (SPIV) measures planar three-dimensional velocity fields through the use of at least two pairs of particle images recorded separately by two different cameras. The measurement accuracy is closely related to the time interval between particle images. Two data sets for a laminar flow field and a turbulent open channel flow field were used to investigate the influence of the time interval on the SPIV measurement accuracy. The results show that when the one-quarter rule is satisfied, the overall measurement accuracy for laminar flow improves with increasing time interval. However, larger time intervals lead to lower measurement accuracies in regions with large velocity gradients when the particle displacement in the interrogation window does not satisfy the two-thirds rule. For the turbulent open channel flow, the measurement accuracy in the outer region changes little with the time interval, but the errors in the near-wall region increase greatly with increasing time interval when the two-thirds rule is not satisfied. For both flows, the measurement errors increase when the one-quarter rule is not satisfied. Thus, the one-quarter rule must always be satisfied and a relatively long time interval will improve the measurement accuracy for flows with negligible velocity gradients. Finally, the longest time interval satisfying the two-thirds rule is the optimum choice for measuring flows with large velocity gradients.

Compliance checking of building designs is essential for providing proper designs for green, safe, and comfortable buildings. However, traditional compliance checking approaches are highly dependent on experienced experts. The procedures are tedious and time-consuming with mistakes and inconsistencies often occurring during the checking process. This paper reviews recent research on automatic compliance checking to reflect the state-of-the-art in this area. A framework for automatic compliance checking was developed with the tasks divided into rule extraction and representation, information modeling and extension, rule reasoning and execution, and result reporting and visualization. These four aspects are used to classify the current status and limitations of the various current methods with the results showing that current information models have many differences with huge semantic gaps that require flexible, unified information models, also further investigations are needed of automatic rule extraction methods with an open environment for rule sharing and improved methods will involve complex spatial relationship and large reason capacities.

Three-dimensional radiation fields must be known to predict the distribution of real external exposures in nuclear facilities. Such data fields provide the basis for establishing effective shielding measures and determining reasonable operating plans at nuclear facilities. Three-dimensional gamma radiation fields were reconstructed using point-kernel integral theory and a source activity inversion algorithm using the least squares method and a Gauss-Seidel iterative algorithm. The results were then used to study the influences of the dose measurement positions on the source inversion results. The reconstruction method was verified using measured data from a nuclear power plant. The results show about 10% difference between the reconstructed gamma radiation field data and the measured data which is sufficient for radiation protection studies of nuclear facilities and for optimizing radiation protection procedures.