To investigate the effects of solar irradiation and wetting-drying cycles on the crack evolution of compacted loess and clarify the underlying mechanisms, laboratory tests were conducted using a xenon lamp to simulate solar irradiation under varying irradiance levels, dry densities, and wetting-drying cycles. Surface crack images were periodically captured using a self-developed acquisition system. Crack morphological parameters were extracted using the Particle and Crack Analysis System (PCAS), and micro-pore structures were quantitatively analyzed based on scanning electron microscopy (SEM) images, enabling a systematic macro-micro analysis of crack evolution characteristics. Results indicate that increasing irradiance accelerates crack initiation and increases crack ratio, main crack length, and overall fractal dimension. Within the dry density range of 1.5-1.7 g/cm³, higher dry density effectively reduces crack ratio and connectivity, thereby inhibiting crack propagation. Under wetting-drying cycles, porosity generally increases, pore circularity decreases, and fractal dimension shows an initial increase followed by fluctuations, corresponding well with macroscopic crack evolution. Solar irradiation enhances surface evaporation, intensifies moisture migration and deformation heterogeneity, and promotes the transition from pore structure adjustment to macroscopic crack propagation.

In order to reduce accident risks caused by leakage during the refueling process, the hydrogen leakage and diffusion behavior of a 35 MPa hydrogen dispenser was numerically simulated using Ansys Fluent. The characteristics of hydrogen leakage and diffusion under the canopy structure in the refueling zone were investigated. The effects of leakage diameters, ambient wind velocity, and local ventilation on hydrogen concentration distribution and the evolution of flammable areas were analyzed. The results show that when the leakage diameter of filling hose is 2 mm, no flammable area is formed on the underside of canopy. However, when the leakage diameter increases to 5 mm and 10 mm, a flammable area can develop on canopy underside. The location of the highest hydrogen concentration on underside of canopy is concentrated near the axis parallel to the jet direction. Specifically, when leakage diameter is 10 mm, the hydrogen concentration on underside of canopy along vertical leakage direction exhibits a Gaussian distribution. When the ambient wind is perpendicular to leakage direction, wind velocities of 2 m/s and 8 m/s can effectively reduce hydrogen accumulation near the leakage hydrogen dispenser. In contrast, at the wind speed of 5 m/s, a vortex structure was formed near obstacles, leading to hydrogen accumulation and increasing the risk of fire and explosion. Under no ambient wind conditions, local ventilation is provided in the refueling zone. When the ventilation velocity reaches 5 m/s and 10 m/s, the hydrogen cloud concentration within the flow field can be successfully diluted to below the flammable limit within 2 s. Moreover, a ventilation velocity of 10 m/s shows a more pronounced effect in reducing the hydrogen concentrations in front of the leakage source.

To enhance China's work safety governance, policy texts related to work safety were analyzed using text analysis and coding methods, aiming to provide an in-depth understanding of the current policy system and to identify existing gaps in China's safety governance policies. Based on Duxiu database and the website of the Central People's Government, 80 policy texts issued between 1949 and 2024 were selected. NVivo 12 plus software was used to conduct coding analysis, which was then employed to summarize the distribution and usage of policy tools in China's work safety governance. The research show that China's work safety governance policies employ a comprehensive set of policy tools. They exhibit a development pattern centered on pre-accident prevention, supported by talents and technology, and complemented by coordinated measures during and after accidents. At the same time, the current policy structure is still unbalanced. Policy tools play a prominent role, while enabling and incentive tools are insufficient. The integration degree of policy tools with governance logic elements is low, affecting the overall effectiveness of governance. In the future, it is necessary to promote the optimization of policy tool structure, balance the distribution of governance elements, and strengthen the deep integration of policies and elements to enhance the level of work safety governance.

To address the problems of low accuracy and insufficient adaptability in existing methods for determining the parameters of PIM for predicting surface deformation prediction in goaf areas under thick unconsolidated layers, 36 sets of measured surface movement data from coal mining working faces were selected. The core indicators of mining-geological conditions were screened via Hierarchical Cluster Analysis (HCA), Entropy Weight Method(EWM) and Grey Relational Degree (GRD) analysis. Furthermore, the GRNN model was optimized by integrating K-fold cross-validation with the neighborhood perturbation strategy of SAA, and an SAA-GRNN optimization model was constructed for PIM parameter determination. A case study was conducted using 45 sets of data from coal mining working faces with thick unconsolidated layers in the Jining area. The results show that: seven mining-geological condition indicators can be classified into three categories, and five core input indicators were identified screening, namely mining thickness M, coal seam dip angle α, mining depth H, strike mining degree D₃/H, and unconsolidated layer thickness h. The maximum root-mean-squared error (RMSE) of SAA-GRNN model is no more than 0.190 4, the maximum mean absolute error (MAE) is controlled within 0.133 9, the maximum mean absolute percentage error (MAPE) is 0.153 6, and the overall coefficient of determination (R²) is generally above 0.8. Under the same conditions, the prediction errors are greatly reduced compared with those obtained using Back Propagation (BP) neural network and the conventional GRNN model.

With the global transition of civil aviation navigation systems from magnetic north to true north reference, a dynamic risk assessment model integrating STPA and FBN was proposed to quantify, identify, and effectively control systemic risks induced by the navigation reference transition. A four-level control structure-covering strategic, regional, organizational, and equipment layers-was established to identify seven categories of system-level hazards and twelve types of unsafe control actions. Expert uncertainty was quantified via fuzzy sets, and a Bayesian network (BN) was constructed using the Leaky Noisy-or Gate model. Furthermore, a dynamic Bayesian network (DBN) was developed to simulate risk evolution across five phases (t₀ to t₀+28 years). The results show that technological lag and insufficient policy coordination are the major risk drivers in the early stage (e.g., airspace conflict probability up to 0.852). However, through phased implementation of technology upgrades, policy alignment, and redundancy design, key risks can be reduced to below 0.01 by t₀+28. This study proposes an original strategy integrating the 'phased compliance-fund disbursement' policy linkage mechanism, aircraft service life-based technical iteration path, and the 'inertial navigation + low-orbit satellite' dual-redundancy artificial intelligence (AI) governance system, to systematically resolve policy delays, intergenerational equipment conflicts and operational risks in the true north transition.

To address the issue of insufficient small-object detection accuracy in remote monitoring of heavy industrial workshops, an unsafe behavior detection algorithm based on improved YOLOv7 was proposed. First, the traditional upsampling was replaced with a lightweight content-aware reassembly of features (CARAFE) module, which effectively preserved the semantic information of small objects through adaptive feature reassembly. Second, an improved Bi-level routing efficient layer aggregation network(Bi-ELAN) module was proposed by integrating the BiFormer dynamic sparse attention mechanism into the head network, which strengthened the multi-scale feature fusion capabilities and established target-background contextual relationships. Third, the loss function was refined by introducing the shape intersection over union(ShapeIoU)loss function, which enhanced bounding box regression accuracy through geometric shape constraints. Finally, ablation experiments and comparative experiments were conducted on the improved YOLOv7 model based on constructed remote monitoring perspective dataset. The results show that, while maintaining model lightweight characteristics, the proposed algorithm significantly improves small-object detection accuracy in remote monitoring scenarios. The improved model achieves a precision of 84.2%, a recall of 78.6%, and a mean average precision (mAP@0.5) of 78.8%. Compared to the original YOLOv7 algorithm, the improved algorithm increases precision, recall, and mAP@0.5 by 5%, 0.3%, and 2.6%, respectively.

To solve the problems of large error of layout parameters and low extraction efficiency of artificially designed high-level gas extraction boreholes, a high-gas mine in Xinjiang was taken as the research object. A design method for pressure-relief gas high-level extraction boreholes based on two-dimensional physical similarity simulation and an intelligent system was proposed. Through the two-dimensional physical similarity simulation test, the evolution characteristics of the horizontal and vertical fractures of the overlying rock were revealed. Additionally, the geometric boundary between the gas migration area (maximum height 36.7 m, maximum width 22.7 m) and the reservoir area (maximum height 26 m, maximum width 17 m) was accurately divided, and the spatial evolution characteristics of gas occurrence were clarified. Based on Python language, the intelligent system of high-level gas extraction borehole was developed, and the 3D geological model is constructed by integrating OpenGL technology. Combined with the parameters such as the horizontal distance between the borehole end point and the opening point, the azimuth angle and the final hole height, the borehole layout parameters (azimuth angle, inclination angle and length) were automatically generated by the self-developed parameter calculation system. Subsequently, the borehole trajectory was simulated by the visual demonstration system. It is shown by the application that the final hole position of the borehole designed by this system is accurately located in the upper part of the caving zone and the middle and lower part of the fracture zone. The gas extraction concentration of 2 # drilling field is recorded at 6.52%—10.94%, which is found to be 2.52%-5.19% higher than that achieved by the traditional method

To effectively reduce the contagion risks in the "last-mile" of emergency logistics in epidemic-stricken areas, a truck-drone collaborative delivery mode was first designed. A "basic reproduction number" function was constructed based on epidemic transmission dynamics to quantify the number of infections at various demand points. Then, a routing optimization model for truck-drone collaborative emergency supply delivery was established, aiming to minimize both the total number of infections and the total delivery time. In view of the multi-objective and non-linear characteristics of the model, the IMOABCA was developed. Finally, experiments were carried out through multiple types of instances. The results show that the IMOABCA could scientifically optimize delivery routes by integrating epidemic data, demand point distribution, and population size. Compared with the basic multi-objective artificial bee colony algorithm (MOABC)and Non-dominated Sorting Genetic Algorithm-II(NSGA-II), the total number of infections is reduced by 922 and 746, respectively. Additionally, the total delivery time is saved by 3.71% and 1.41%, and the task completion time can be shortened by 14.06% and 3.6%, respectively.

To solve the limitations of traditional fault trees in accurately capturing the complex correlations among components of communication base station systems, this study proposes a method for seismic fragility and importance analysis based on T-S fault tree. First, a three-subsystem architecture consisting of power supply, machine room, and transmission is constructed, and a T-S fault tree model for post-earthquake functional loss of communication base station systems is established. The functional correlations among components and subsystems are quantified using gate rule tables. Second, the seismic fragility models of communication base station system and the machine room subsystem are comparatively analyzed. Then, the seismic fragility of communication base station system is calculated using traditional fault trees and Monte Carlo simulation, and compared with the results from T-S fault tree. Finally, key impact factors are identified by combining the T-S critical importance analysis, and the weak links of the system are located through the sensitivity analysis of component fragility parameters. The results show that analyzing only the seismic fragility of the machine room subsystem underestimates the risk of system functional loss, and it is necessary to comprehensively analyze the seismic fragility of the communication base station system by integrating the three subsystems. The T-S fault tree method has advantages in describing the fuzzy logic relationships among components, and its results are more reliable than those from traditional fault tree. Cables under the slight damage state and machine room buildings under the severe damage state are the key impact factors. Substations, transmission lines, and machine room buildings have the most significant impact on the system function.

In order to address the challenges of quantitatively evaluating public behavioral responses to rainstorm disasters and clarifying the degree of their alignment with disaster risks, this study took the Shenzhen "9·7" rainstorm as a case study. Employing the PSR model and public LBS data, public response behaviors during the disaster were comprehensively evaluated from three dimensions: pressure, state, and response. A response adaptation index was established to measure the alignment between public behavior and rainstorm disaster risks. The findings indicate that, compared to normal conditions, the public's travel patterns during rainstorms exhibit similar spatial characteristics but with reduced intensity. Across all time phases, the response effectiveness is highest during non-peak daytime hours. Among different administrative districts, residents in Futian and Longhua District demonstrate the highest level of responsiveness, while those in Yantian District exhibit a relatively weaker response. Among various functional area types, schools and recreational areas show the most significant reduction in travel intensity, indicating the most positive public response, whereas residential and office areas showed a comparatively weaker response.