In order to improve the inspection efficiency and safety of UAV inspections for wind turbines，a reasonable planning of the UAV inspection path was proposed. A method for UAV inspection path planning based on real-time emergency landing safety constraints was introduced. First，a safety calculation model for emergency landing areas was established. It based on the dynamic endurance capacity of UAV affected by wind speed and direction，as well as constraints such as flight path emergency landing，to assess the safety of the inspection path and establish safety constraints. Then，regarding the objective function of the length of UAV inspection path，an optimal inspection path planning method based on the characteristics of RSA was proposed. This method effectively solved TSP for UAV inspections in wind farms under safety constraints，utilizing the discrete and multi-agent characteristics of RSA algorithm to plan the inspection path for wind farms. Finally，comparative experiments and simulations of wind farms were conducted for different algorithms. results indicated that the real-time emergency landing safety constraint model can comprehensively calculate safe routes by integrating various risk factors，enhancing the safety of the inspection path. RSA algorithm can quickly solve TSP problem for wind farm inspections under safety constraints，improving the level of inspection path planning.

In order to elucidate the specific causes of abnormal alarms from CH₄ and CO sensors in the return corner of the backfill working face，a systematic investigation was conducted. Initially，a programmed heating-gas chromatography (GC) experiment was carried out on filling materials，complemented by on-site GC measurements，to evaluate whether the alarms were attributable to CH₄ and CO concentrations exceeding threshold limits. Subsequently，a portable gas detector was employed to monitor various filling materials，identifying the primary materials responsible for triggering the sensor alarms. Finally，GC-mass spectrometry (MS) analysis was performed to characterize the volatile components of adhesives and their interference effects on CH₄ and CO sensors. The results indicate that the alarms triggered by CH₄ and CO sensors were caused by the volatile gases from adhesives，rather than by an excessive concentration of CH₄ or CO. The primary constituents of the adhesive VOCs were alkanes，while secondary components included alcohols and esters. Key interfering substances for CH₄ sensor were alkanes such as C₅H₁₂，C₆H₁₂，and C₆H₁₄，with minor contributions from alcohols and esters such as CH₄O，C₂H₄O₂，and C₃H₈O₂. All ten tested combustible gases exhibited cross-interference effects on CH₄ sensor. Interfering substances for CO sensor included CH₄O，C₂H₄O₂，and C₃H₈O₂. While the sensors demonstrated short-term resilience to interference under abnormal gas atmospheres，their stability and anti-interference performance significantly deteriorated with prolonged exposure.

In order to assess the resilience of CPS in airport flight areas and provide a reference for rapid recovery in emergencies，a refined and real-time airport flight area CPS network model with flight area control network as the information network and flight area taxi path network as the physical network was proposed. For CPS network in airfield areas，the connectivity of the network was calculated by selecting the relative values of the largest connected subgraphs. The network's resilience was evaluated by combining robustness，performance loss，and comprehensive resilience indicators. CPS situation in airfield areas under different disturbance recovery strategies was compared to determine the optimal recovery strategy. Taking Xi'an Xianyang Airport flight area CPS as an example，the results show that betweenness perturbation causes the greatest damage to the control network，while degree value perturbation causes the greatest damage to the taxiing path network. Using betweenness recovery can enable faster recovery of CPS resilience in airfield areas，and CPS network of the flight area of the airport shows a higher level of resilience under random perturbation. The research results can simulate and predict relevant nodes and provide certain reference significance for ensuring the safety of flight area operations.

To reveal the impact of coal metamorphism on the self-ignition process，experimental and quantitative analysis methods was combined to analyze the pyrolysis characteristics，microstructural changes，and their correlations during coal self-ignition. First，Fourier-transform infrared spectroscopy (FTIR) was employed to determine changes in the content of active functional groups in coal samples. Then，thermogravimetric analysis (TGA) was conducted to obtain pyrolysis characteristic parameters for coal samples with varying degrees of metamorphism. Finally，Pearson correlation coefficient method was applied to quantify the relationship between microstructure and characteristic temperatures during coal self-ignition. The results indicate that as the degree of coal metamorphism increases，the content of aromatic and aliphatic hydrocarbons(Ar-CH) rises，while oxygen-containing functional groups(C-O-C) significantly decrease. Low-rank coal samples exhibit the highest proportion of oxygen-containing functional groups，whereas high-rank coal samples contain more than 50% aromatic hydrocarbons. Increased metamorphism delays weight loss curves and shifts them to higher temperature regions，leading to an increase in characteristic temperatures. Ar-CH show a significant positive correlation with characteristic temperatures，while C-O-C exhibit a negative correlation. Overall，the increase in Ar-CH and the decrease in C-O-C in coal samples are identified as the primary microstructural factors responsible for the rise in characteristic temperatures and the reduction in self-ignition propensity.

In order to address the issue of missed detections during small target detection of aircraft rivets，an improved YOLOv8n algorithm for the detection of aircraft rivets and their anomalies was proposed. First，by adding a small object detection head，the shallow detail information in the backbone network was better fused，enhancing the model's feature fusion capability and its ability to recognize and locate small rivet targets. Second，the first two convolutions in the backbone network were replaced with SPD-Conv，which reduces information loss during down sampling through the combination of feature map reorganization and non-stride convolutions. Finally，large separable kernel attention (LSKA) was integrated into the spatial pyramid pooling fast (SPPF) module，capturing the dependencies between spatial and channel dimensions by calculating spatial and channel weights on each feature map and adjusting the feature maps to enhance the algorithm's ability to extract and recognize rivet feature information. Ablation experiments and comparative experiments were conducted based on a self-built aircraft rivet dataset. The results show that the proposed algorithm can achieve real-time identification of aircraft rivets and their anomalies，with precision，recall，and mean average precision (mAP) values improved by 6.5%，16%，and 15%，respectively，compared to the YOLOv8n algorithm. The detection performance is also significantly better than other mainstream algorithms.

In order to improve the safety production level of coal mining enterprises and ensure a reasonable proportion of safety investment，the CPT method was introduced to establish a dynamic safety investment decision-making model that combined the criteria importance through intercriteria correlation(CRITIC) method improved by the coefficient of variation method and CPT. From the perspective of safety input and output，an evaluation index system was constructed，and the weights of each index were calculated by the improved CRITIC method. Combined with the CPT，the static decision-making ranking results of various safety investment schemes in coal mining enterprises were calculated，and on this basis，the results were dynamically optimized to obtain the final dynamic decision-making ranking results of safety investment. The study shows that the constructed model is reasonable and reliable with high sensitivity compared with VlseKriterijumska Optimizacija I Kompromisno Resenje (VIKOR) method. The model considers the influence of the decision maker's subjective risk preference on the actual decision-making results，and the time factor can reflect the dynamic accumulation level during the research period，so the decision-making results are more scientific and reasonable.

To accurately identify and evaluate the fire risk in large-scale urban complexes，a fire risk assessment indicators system was constructed，consisting of 32 secondary indicators across six dimensions: building fire protection，fire protection facilities，electrical fire prevention，fire safety management，emergency response and coordination，and fire protection technical services. The order relation method was employed to determine the subjective weights of each indicator，while the entropy weight method was used to calculate the objective weights. Subsequently，a game-theory-based combined weighting approach was utilized to derive integrated weights，and a cloud model was applied to quantitatively analyze the fire risk of a large complex in a specific city. The results demonstrate that this combining model not only mitigates the subjectivity limitations in indicator weighting but also accounts for the fuzziness and randomness in the evaluation process. The primary risks of the large complex stem from fire protection technical services as well as emergency response and coordination，requiring preventive measures. Furthermore，the evaluation results are consistent with the actual situation，validating the feasibility of the proposed model in assessing fire risks for large-scale urban complexes.

In order to reduce the failure risk of petrochemical storage tanks in coastal areas under the influence of hurricanes，the failure model for vertical cylindrical tanks under wind load was established based on the strength stress interference theory. The critical wind speed for the buckling failure was determined by an example of a crude oil storage tank. For different random distributions of parameters，such as wind speed，the Monte Carlo method was also used to plot the probability curve of shell buckling failure. The buckling failure mechanism under static conditions was investigated，and the failure probability at different loading levels was determined under static conditions. The results show that the critical wind speeds are 60.67，65.38 and 69.79 m/s respectively when the loading levels are 25%，50% and 75% under the same wind load. According to the failure probability curve，the failure probability of the loading level of 25% is significant greater than that at 50%. Therefore，it can be seen that the buckling failure mechanism of the vertical cylindrical storage tank shell is as follows: the loading level is too low，which can easily lead to buckling failure of the storage tank under wind load. The higher the loading level，and the stronger the wind load resistance of the vertical cylindrical storage tank. Appropriate measures can be taken to increase the loading level to resist strong wind loads in the coastal chemical industries.

In order to strengthen the governance of online public opinion of chemical emergencies and properly handle the online public opinion crisis caused by chemical emergencies，evolutionary game theory was introduced into the process of network public opinion governance，and a binary evolutionary game model was constructed for local government and network media. Combined with the SD model，a quantitative analysis model was constructed for local governments and online media. Simulation research was conducted based on relevant cases，and the strategic evolution process of each game subject was compared and analyzed. The results indicate that the popularity of online public opinion on chemical emergencies depends on the strategic choices of each party，and the evolutionary game model analysis shows a periodic and recurrent trend. After introducing a punishment mechanism，appropriately increasing the severity of punishment can bring the evolutionary game system into a benign state. Local governments can enhance the emergency warning mechanism for online public opinion，strengthen daily supervision and collaborative governance of online media，and formulate reasonable punishment measures to effectively prevent local governments from inaction and disorderly behavior in the process of responding to chemical emergencies and online public opinion. This can achieve supervision of online media and avoid the vicious evolution of online public opinion.

In order to further scientifically prevent and control coal mine gas accidents and systematically analyze the risk factors and coupling relationships of coal mine gas accidents in my country，an association rule mining model based on the PSO algorithm using Python software was established and verified. The risk factors of coal mine gas accidents were classified in combination with the HFACS accident risk model，and the constructed PSO-FP(Freguent Pattern)-growth algorithm was used to mine association rules for coal mine gas accident investigation reports. The results show that the PSO-FP-growth algorithm has better running speed and association rule effect than the PSO-Apriori algorithm. According to the visualization of association rules of gas accident risk factors and high-support association factors，the main risk factors for coal mine gas accidents in my country are defects in safety supervision and management of coal mine enterprises，inadequate gas prevention and control technology，weak safety awareness of employees，and inadequate management awareness and technology of on-site managers.