Abstract Fault activation serves as a key mechanical mechanism inducing dynamic hazards such as rockbursts and mine tremors during deep coal seam mining. Its evolution is significantly influenced by disturbances from overlying goaf areas and the advancement method of the working face. Taking the F17 normal fault in the South 2 Lower Mining Area of Dongrong No. 2 Mine as the engineering context, a three-dimensional numerical model was constructed under the condition of an existing upper 16# coal seam goaf. Comparative analysis was conducted for two working face advancement scenarios along the fault’s hanging wall and footwall in the 17# coal seam. The evolution characteristics of fault stability were characterized using the ratio of shear stress to normal stress on the fault plane. Results indicate that the upper goaf significantly reduces normal stress on the fault’s hanging wall, triggering shear response within approximately 20–40 m of the working face. Under upper-plate advancement conditions, the shear stress/normal stress ratio gradually increases and stabilizes within the 0.20–0.25 range, with fault activation manifesting as progressive slip and a maximum relative displacement of approximately 120 mm. During the lower plate advance, shear stress rapidly concentrates, causing the shear stress/normal stress ratio at the fault plane to exceed 0.25. Fault slip exhibits sudden intensification, with a significantly larger affected area and higher intensity compared to the upper plate advance. Research indicates that the combined effects of the weakening influence of the goaf and the superposition of mining-induced stresses jointly govern the asymmetric evolution of fault activation. Specifically, the upper plate advance is primarily characterized by gradual activation, whereas the lower plate advance is prone to inducing sudden instability.

Abstract The crack compliance method is widely utilized in the initial residual stress testing of aviation aluminum alloys due to its broad applicability to various materials and specimen sizes, high solution accuracy, and minimal additional error. The test method of the crack compliance method is analyzed to assess the influence of strain test error on the accuracy of stress calculations; A quantitative analysis model is constructed to evaluate the relationship between stress and strain, determining the sensitivity of strain error on stress. Research indicates that the ill-conditioning (high condition number) of the stress–strain relationship matrix is the root cause of error amplification. By introducing elementary transformations to preprocess the transfer matrix, its numerical properties can be effectively improved, and its condition number can be significantly reduced. Deformation experiments and finite element simulations demonstrate that the relative error between the stress solution model preprocessed with elementary transformations and the actual deformation amount is reduced by 14.48% compared to the traditional model. This indicates that this method can effectively improve the accuracy of deformation prediction. This study offers a theoretical foundation and feasibility proof for stress testing and deformation control within the aerospace manufacturing sector.

Abstract Electric vehicles (EVs) are becoming essential transportation alternatives due to their pollution-free environment. Researchers use EV power train models to study battery monitoring, range estimation, and vehicle performance. The models are simulated using various drive cycles and use an IoT-based model to send vehicle data to the cloud for communication optimization. The Li-ion battery pack is used for the EV battery model. The model assesses the EV’s full powertrain and mathematical model, determining efficiency, motor output, and battery parameters as a function of torque and speed. Motor torque is crucial for a vehicle’s power consumption and mobility. Greater mass but higher SOC consumes more energy over the same distance—a higher wind speed results in 6.369 kWh of energy consumption and a balance of 61.52%. The proposed system tracks user operation data to train a remaining distance prediction model based on an electric car’s battery state of charge. The simulation outcomes summarize the vehicle’s distance travelled and the battery state of charge.

Abstract Artificial electric field algorithm (AEFA) is an effective approach for tackling diverse optimization problems. However, it is prone to premature convergence and exhibits high computational complexity in handling complex optimization problems. To alleviate these deficiencies, a hybrid AEFA with multi-population collaboration (MCHAEFA) is proposed. Specifically, in MCHAEFA, the entire population is first divided into five subpopulations of the same size. On this basis, a leader-adaptive strategy is introduced to achieve communication among subpopulations. Subsequently, the original position updating equation is hybridized with the exploration strategy from the recently developed snow ablation optimizer (SAO) to strengthen global search. Finally, a novel attenuation coefficient is adopted to better balance exploration and exploitation. To investigate the effectiveness, MCHAEFA is compared with five existing AEFAs and 35 well-known optimizers on 29 CEC 2017 benchmark functions. Moreover, MCHAEFA is applied to multi-robot path planning. Experimental results indicate that the presented MCHAEFA is highly efficient and competitive.

Abstract This paper proposes a high-precision lane detection and correction method based on roadside millimeter-wave radar. After deploying millimeter-wave radars on traffic light poles to collect lane point clouds, spatiotemporal adaptive DBSCAN clustering is employed to extract lane clusters, with adaptive neighborhood and temporal weighting improving clustering accuracy. To compensate for sensor offsets, an ICP algorithm with curvature keypoint selection is introduced for rigid alignment, enhancing correction stability. Lane shapes are fitted using a segmental sliding-window PCA combined with convex hull calculation and B-spline interpolation, achieving high-precision modeling for complex curves and nonlinear lanes. Experimental results demonstrate that the proposed method effectively handles environmental variations and noise, significantly improving the accuracy and robustness of lane detection and correction.