In order to reveal the complex causality between EWSV and their multiple antecedent conditions，and to improve the efficiency of safety governance，a comprehensive model integrating contemporary deterrence theory，protection motivation theory，and social learning theory was constructed from a perspective of complexity theory. Based on this，six antecedent conditions affecting EWSV were identified from three perspectives: leader，coworker，and employee. Then，the fsQCA was used to reveal what configuration of antecedent conditions would lead to high level of EWSV. The results show that a single antecedent condition is insufficient to explain high level of EWSV but safety-specific leader punishment omission and coworker work safety violations(CWSV) play universal roles in forming high level of EWSV. Three types of driving modes composed of five condition configurations can lead to high level of EWSV. Three types of condition configurations lead to non-high level of EWSV. Reducing CWSV and improving employees' perception for formal sanctions are crucial for achieving non-high level of EWSV. Different combinations of multiple antecedent conditions can lead to high level of EWSV，and there is a complex causality (concurrency，equivalence，and asymmetry) between high level of EWSV and their antecedent conditions.

In order to prevent high-level fire hazards such as spontaneous combustion of top coal and upper layer residual coal，a self-produced gas expansion slurry was prepared by taking industrial solid waste phosphogypsum as the base material，adding components such as sodium bicarbonate and polyacrylamide. Firstly，a thermogravimetry(TG) analyzer was used to study the thermal stability of the slurry at different heating rates. Then，a self-built insulation performance testing system was used to test the thermal insulation performance of the slurry. Finally，the fire extinguishing and cooling performance of the slurry was studied through fire extinguishing experiments. The results show that when the temperature of phosphogypsum based self-produced gas expansion slurry is below 160 ℃，there is a mass loss of 35% to 46% in the slurry. When the temperature is above 160 ℃，the mass fraction of the slurry no longer changes. The thickness of the slurry accumulation reaches 30 mm，and the effective insulation time exceeds 139 min，which can achieve good coverage of coal spontaneous combustion hazardous areas. The slurry expands and diffuses in the cracks of the coal pile，completely enveloping the broken coal from bottom to top，isolating oxygen，and having good cooling and fire extinguishing performance. After 36 min of grouting，the temperature at all points has been decreased to below 50 ℃，and no reignition phenomenon has occurred.

To investigate the development trend of occupational health of coal miners and promote the implementation of occupational health protection in coal mines. Based on the knowledge graph analysis method，CiteSpace software was used to conduct the keyword clustering and emergence of coal miners' occupational health literature included in the core databases of China National Knowledge Infrastructure(CNKI) and Web of Science(WoS) in the past 20 years. The results indicated that coal miners' occupational health research topics were divided into occupational health，occupational hazard factors，and career-related diseases. Occupational-related diseases focused on behavioral and physical diseases，cardiovascular and cerebrovascular diseases metabolic diseases，and chronic non-specific respiratory diseases. Compared with occupational diseases，occupational-related diseases had the characteristics of 'multiple causes and one effect'，causing much more difficulty in diagnosis. Pneumoconiosis，hearing loss，psychological stress，and hypertension were the main factors threatening coal miners' health. Moreover，such health issues were associated with occupational exposure，and risk analysis and control were important strategies to deal with hazard factors. The emerging research hotspots were the cause and prevention of occupational-related diseases，and the multi-discipline research of occupational health，brain science，and intelligence.

In order to improve the safety level of China Civil Aviation，a cause analysis method，which focused on the whole process of 'taxiing-takeoff-climbing-cruising-falling-approaching and landing'，was proposed based on association rules and SNA. Firstly，a total of 1 051 civil aviation accident cases from 2010 to 2022 were collected from Aviation Safety Information System. Then，a civil aviation accident causation index system based on 'human-machine-environment-management' was established in combination with 'control navigation service procedure-training'. Secondly，the EasyData data service platform was used for annotation，followed by constructing the data-set of the cause of civil aviation accidents facing the whole process of flight. Furthermore，the FP-Growth algorithm was employed to mine the association rules with the flight phase as the subsequent term of rules. Finally，this study emphasized the co-occurrence frequency of association rules，and developed the correlation network of civil aviation accident causes by applying the centrality analysis of flight phase and accident causes. The results show that the most dangerous phases of flight are the approach and landing. Operation failures，runway safe accidents and cabin safe accidents are the main types. Among the 10 identified causes of civil aviation accidents，there are 4 human factors，2 facilities and equipment factors，3 environmental factors and 1 management factor.

In order to effectively provide early warning of the occurrence trend or state of prefabricated building workers' unsafe behaviors (PBWUBs) of working at height，and to enhance the control of PBWUBs，RF-SFLA-SVM model was proposed to conduct an early warning study on workers' unsafe behaviors. Firstly，the SHEL (Software-Hardware-Environment-Liveware) model was used to analyze the factors influencing the unsafe behaviors of prefabricated building workers in danger of working at height. RF was used to determine the key warning indicators. Then SFLA was used to find the best parameters for SVM. Finally，the RF-SFLA-SVM model was used to predict and warn about the unsafe behavioral state of the prefabricated building workers working at height，and its performance was compared with other warning models. The results show that the RF-SFLA-SVM-based warning accuracy of PBWUBs of working at height was the highest，91.67%，which was a maximum improvement of 14% compared with the warning performance of other models. The research results can give a reference for the control and prevention of PBWUBs working at height.

To advance environmental risk prevention and control in the Yellow River basin and achieve high-quality development，this study utilized air pollution，water pollution and carbon emissions to characterize environmental risk factors in this region. Firstly，the kernel density estimation method was employed to construct a kernel function，perform density smoothing，and analyze the spatial agglomeration patterns and evolutionary characteristics of environmental risk factors. Secondly，a spatial autoregressive model that accounts for spatial dependence was adopted to dissect the influences of multidimensional factors on environmental risk factors. The results indicate that air pollution，water pollution，and carbon emissions are primarily concentrated in the middle and downstream cities of Yellow River basin，delineating distinct regions for environmental risk prevention and control. Optimizing industrial structures and promoting green technological innovation in enterprises can reduce environmental risk factors. However，unreasonable energy conservation and environmental protection expenditures may lead to an increase in environmental risk factors. Furthermore，an inverted U-shaped relationship was observed between economic levels and population density and risk factors. From a long-term strategic perspective，regional collaborative governance and synergistic pollution reduction and carbon mitigation should be the strategic priorities. Strengthening cooperation among middle and downstream cities，optimizing industrial structures and increasing investment in industrial technological innovation are crucial steps towards achieving an economic development model transformation.

To improve the accuracy of miners' helmet-wearing recognition in coal mine underground operations，a helmet-wearing recognition algorithm was proposed based on human posture analysis and machine vision system optimization methods. First，Single Shot MultiBox Detector (SSD) was used as the basic model of multi-target identification. The squeezed neural network (SqueezeNet) was used to reduce the model parameters to develop an efficient recognition model，which improved the recognition accuracy of the miners' helmet and maintained the balance between the recognition accuracy and the calculation speed. Then，a multi-person posture estimation algorithm was used to locate the joint points of each miner and determine the miners' complex behavioral status. Finally，the upper limb nodes of the target were extracted based on fusion model of multi-target recognition and multi-person posture estimation，and then the helmet-wearing condition was determined by the spatial topological relationship between the upper limb nodes and the helmet frame. Moreover，3 000 pieces of image data were selected to validate the proposed method's performance. The results indicated that the machine vision system can optimize the hardware and software configuration to improve the system's performance. Furthermore，the identification accuracy reached up to 91.1%，which was much better than that of the latest helmet-wearing recognition algorithm. Therefore，the proposed system in this study can meet the requirements of helmet-wearing recognition accuracy for underground coal miners.

During the construction of underground extraction boreholes，trajectory deviation is common，prone to blind gas extraction areas and produces gas extraction blank zones. In order to ensure that gas extraction achieves the expected results，the 15121 bottom pumping lane，15124 low level lane and 15124 return air lane of Xinjing mine in Shanxi province were taken as the background of the project. We tested the deviation of borehole drilling through the seam and along the seam respectively，calculated the three-dimensional trajectory of the drill holes by using the mean angle full distance method，established the three-dimensional trajectory of the drill holes and the three-dimensional layer relationship of coal rock based on the geological information of coal seam assignment，realized the visualization of the trajectories of the drill holes，analyzed the factors affecting the deviation of the drill holes and the experimental fitting data，and came up with a characteristic of the deviation of the boreholes. The research findings suggest that cross-layer boreholes initially exhibit deviation within fine sandstone layers，with the deviation intensifying notably within the coal seam segment，reaching a maximum deviation of up to 0.52 meters within this section. In the case of in-seam drilling，greater drilling depth results in larger vertical deviations，with an average vertical offset of 1.37 meters at a depth of 40 meters when the opening angles of boreholes are similar. Overall，low-angle boreholes are more susceptible to gravity-induced deviations，leading to a higher likelihood of trajectory deflection，while high-angle boreholes tend to exhibit vertical deviations，albeit with a smaller overall deviation compared to low-angle boreholes. In the horizontal direction，borehole spacing is reduced by 0.68 meters for bottom drainage galleries，0.28 meters for lower position galleries，and 0.54 meters for return airways. The research results effectively guide the design of subsequent gas extraction drilling trajectories under the same geological conditions，provide theoretical support for controlling and eliminating the blind zone of gas extraction，and improve the efficiency and effectiveness of gas extraction.

In order to study the evacuation of passengers in a train fire in a vacuum tube transportation system and the smoke spreading law，The fire numerical simulation software FDS and the personnel simulation evacuation software Pathfinder were used to study the evacuation of passengers in case of fire in a high-speed train compartment consisting of 5 cars in a low-vacuum tunnel. Comprehensively comparing the evacuation time，the degree of smoke spread and the CO concentration in 10 evacuation schemes，the best evacuation method for passengers was obtained，and a rescue vehicle was designed to assist evacuation by docking the door of the high-speed train. The results show that: when the passengers in the fire compartment are evacuated to the neighboring compartments，the best evacuation method for the passengers is to leave the two rows of passengers close to the doorway and the rows of passengers close to the fire source at the same time，and then evacuate the passengers row by row by the distance from the fire source from near to far. If the rescue vehicle is used to assist the rescue，when the fire occurs in compartment 1，compartment 2，or compartment 3，the total time to complete the evacuation of all passengers after the arrival of the rescue vehicle is 533，586，and 376 s，respectively. When a fire occurs in compartment 1，the utilization time of door 1 is 200 s; when a fire occurs in compartment 2，the utilization time of door 1 is 145 s; when a fire occurs in compartment 3，the utilization rate of the two doors is more balanced. Therefore，during the actual evacuation，passengers can be guided to fully utilize the two doors in the form of voice announcements to save the evacuation time. The research results can provide a reliable reference basis for the emergency rescue of passengers in future low vacuum tunnel train fires.

In order to solve the problem that a large amount of multi-source heterogeneous structured and unstructured knowledge in the operation and management of cable crane is difficult to reuse，a large number of technical specifications，expert experience and other safety knowledge accumulated in the installation and operation and maintenance of cable crane were analyzed，and related entity classes of cable crane operation were extracted. A quintuple model was used to structurally express the safe operation of the ontology knowledge of the cable crane，and the binary relationship between classes was established. Then，the ontology expression language in Protégé was used to represent it programmatically. The ontology knowledge base of the cable crane entity，safety early warning and an emergency plan was constructed. 3D scene model of the cable crane was constructed using 3Dmax software，and then the model was imported into Unity3D engine. Knowledge retrieval of the ontology knowledge base was realized through user interface design and interaction functions in the engine. At the same time，based on the existing examples of faults and accidents，the corresponding emergency plans were associated with the hidden dangers existing in each entity of the cable crane. Finally，a complete 3D visualization platform for safe operation knowledge retrieval of cable cranes was constructed. The results show that all kinds of safe operation knowledge of cable-winch can be centrally and effectively managed by constructing an ontology knowledge base，and a knowledge retrieval platform combined with an ontology knowledge base can effectively realize efficient query and reuse of knowledge.