Flight trajectory prediction plays a crucial role in ensuring safe and efficient air traffic operation. In order to consider the implicit correlations between flight trajectory characteristics，the encoding and decoding operations of the prediction framework in the transformer model were enhanced. Firstly，the convolutional block was improved，and ordinary convolutions were applied to capture the correlations between neighboring time series trajectory characteristics，and dilated convolutions were added to capture correlations between non-neighboring time series trajectory characteristics. Secondly，multi-head self-attention was employed to perform calculation based on the spatiotemporal features of the flight trajectory combined with the importance of attention scores. Thirdly，probabilistic sparse method was designed to reduce the computational complexity of the multi-head self-attention and improve the model's computational efficiency. Finally，an experimental platform was established to verify the flight trajectory prediction framework. The results show that compared to the traditional transformer model and the other three neural network models，the improved transformer model shows a 14.4% improvement in time performance. By using root mean square error(RMSE) and mean absolute error(MAE) as evaluation metrics，the average prediction deviations of the improved transformer model for trajectory features such as longitude，latitude，and altitude are 0.027 and 0.021，respectively. These deviations are reduced by 0.072 and 0.063 compared to the traditional transformer model's average prediction deviations of 0.099 and 0.084. Sensitivity analysis on the lengths of prediction sequences indicates that the improved transformer model is more stable than the baseline models.

In order to explore the factors influencing accident injuries in SRBE，a binary LR model was constructed based on data from 1 411 accidents from a Chinese shipbuilding group. OR were used to quantify the impact of factors such as enterprise，time，location，personnel，environment，and accident types to accident injuries. The result identifies 11 significant factors influencing accident injuries. Non-contract workers face significantly higher risks of injury compared to contract workers. Non-approved hazardous operations pose 3.246 times the risk of accident injuries compared to approved hazardous operations. Male perpetrators present a significantly higher risk of causing accident injuries than female perpetrators. The education level of the workers is predominantly at junior high school，and higher education levels are associated with a lower risk of causing accident injuries compared to those with junior high school education. Accident injuries exhibit seasonal characteristics，with accidents occurring frequently in the second quarter，while relatively fewer and less risky in the first quarter. Peak working hours and workdays significantly increase injury risks compared to off-peak hours and non-working days. Object strikes are the primary type of accidents，and mechanical injuries significantly increase the risk of accident injuries. Ships and workshops are the most likely locations for accident injuries within the shipyards. Longer years of service reduce the risk of accident injuries，while higher temperatures increase it.

To address the classification and quantitative evaluation issues of unsafe behaviors of civil aviation pilots，the management mode of their unsafe behaviors was optimized. A method for managing unsafe behaviors in civil aviation was proposed focusing on the civil aviation flight field. Firstly，based on unsafe behavior classification theory，intervention and improvement methods for errors and violations were systematically analyzed to distinguish between errors and violations. Secondly，expert interviews and questionnaire surveys were used to propose a classification evaluation index system for pilots' unsafe behaviors. Furthermore，a classification method of unsafe behavior based on quick access recorder (QAR) data and a quantification method of unsafe behavior scores based on flight operations quality assurance (FOQA) were proposed to achieve a classified quantitative assessment of pilots' unsafe behaviors. Finally，the classifying and managing unsafe behavior process was analyzed and validated. The results indicated that the 68 FOQA monitoring events obtained by screening and calculation had different causal behavioral tendencies. The two proposed unsafe behavior classifications and quantitative evaluation methods can be combined with QAR data or FOQA records. Moreover，the classification management of unsafe behaviors of civil aviation pilots can be achieved by combining intervention and improvement measures for errors and violations.

The construction speed and number of vehicles passing through multi-portal underpass tunnels are increasing，and special driving environments such as long downhill slopes at the portals and underground merging areas cause frequent traffic accidents. To deeply analyze the variation characteristics of lateral offset in tunnel special sections，a test vehicle equipped with inertial navigation and Mobileye was used to perform a real-car natural driving test in a typical multi-entry underpass tunnel. Based on the tunnel's alignment and spatial variation characteristics，the test section was divided into the external tunnel section，the downhill section，the internal section，the underground merging section，and the internal tunnel section. The results indicated that the vehicle trajectories in the other four sections were more complex than the external section of the tunnel，and the lateral offset increased significantly，reaching up to 1.888 to 2.184 times the external section of the tunnel. The lane offset variation rate of the underground merging section and the inner section of the entrance was the smallest. The land discreteness of the underground merging section in the tunnel was the largest，and the discreteness of the entrance downhill section was the smallest with a standard deviation of only 0.111. The average width of the predicted interval of the entrance downhill and underground merging sections were 0.2 m and 0.24 m，respectively. Therefore，the driving safety hazards of the entrance downhill and underground merging sections were higher than other sections.

In order to prevent accidents caused by the failure of the flange sealing groove in the hydrogenation reactor，failure analysis methods such as macroscopic inspection，fracture analysis，ferrite detection，chemical composition analysis，hardness detection，metallographic detection，SEM analysis and operation process analysis were adopted. The influencing factors of damage mode，start-stop operation process，ferrite content in the weld overlay layer，material properties，and abnormal elements were studied，and the reasons for the failure of the flange sealing groove at the outlet of the hydrogenation reactor were analyzed. The results indicate that the failure of the sealing groove is mainly due to the high stress between the weld overlay layer and base metal of the sealing groove，which produces stress corrosion cracking under the action of a corrosive medium containing F，S and other elements. There is no weld overlay layer on the bottom surface of the cracked flange sealing groove，and the transgranular cracks started at the junction of the surface weld overlay layer and the non-weld overlay layer，mainly on the surface of the non-weld overlay layer side. According to the failure causes，the corresponding improvement measures are put forward from the aspects of manufacturing，material selection and maintenance.

In order to accurately and quickly achieve relation extraction from few-shot emergency plan texts，KMKP based on knowledge prompts was proposed. First，a prompt template was constructed，utilizing learnable typed entity markers that incorporate relation semantics. The effectiveness of input guidance on the pre-trained language model (PLM) was thereby enhanced by these markers. Second，the boundary loss function was utilized to optimize model training，enabling the PLM to learn specific dependency relationships in the emergency domain and apply structured constraints to [MASK] predictions. Third，a gradient-free emergency knowledge storage database was created using the training data，and a knowledge retrieval mechanism was constructed by integrating KNN algorithm. The feature connections between training and prediction data can be captured through this mechanism and the gradient-free normation was used to correct the predictions of PLM. Finally，the experimental validation and analysis were performed using four public datasets under few-shot settings (1-，8-，and 16-shot). The results show that compared to the state-of-the-art model，KnowPrompt，F1 score is boosted by an average of 2.1%，2.8%，and 1.9% by KMKP. In a 16-shot emergency plan instance test，a relation extraction accuracy of 91.02% is achieved by KMKP. Catastrophic forgetting and overfitting issues in few-shot scenarios are effectively mitigated.

To eliminate the impact of complexity and uncertainty of safety risks in mountainous scenic areas on operational safety，a risk assessment method for mountainous scenic areas was proposed. Firstly，risk factors in mountainous scenic areas were identified to develop a risk assessment index system including personnel，equipment and facilities，environment，and management. Then，FBN and AHP models were proposed to evaluate risk probabilities and losses. Moreover，an improved ALARP criterion was used to analyze the comprehensive safety risk of mountainous scenic areas. Finally，the performance and effectiveness of the risk assessment method were validated against safety risk assessment in mountainous scenic areas in Beijing. The results indicated that the BN-based risk assessment method for mountainous scenic areas effectively addressed the issue of complex risk factors and interdependent relationships between each level. The combination of BN and triangular fuzzy number can make full use of expert experience and avoid the subjectivity of expert opinions to a certain extent. The key risk factors in mountainous scenic areas were inadequate detection of dangerous amusement facilities，insufficient configuration or arrangement of forest fire prevention facilities，inadequate protective fencing for hazardous amusement projects，and rockfalls and landslides.

To address the issue of predicting heat stress risks for police officers engaged in outdoor operations under high-temperature conditions，a test dataset for monitoring core temperature of police officers under different environmental working conditions，levels of physical exertion and clothing scenarios was constructed. First，features such as height，weight，age，gender，body fat percentage，physical activity ratio (PAR)，clothing insulation (CI)，environmental temperature and relative humidity were extracted. Then，machine learning methods，including K-nearest neighbors (KNN)，random forest (RF) and gradient boosting decision trees (GBDT)，were used to establish predictive models of core temperature and heat stress risk for outdoor police officers. These models were subsequently validated. The results indicate that for the predictive model of core temperature for outdoor police officers working in high-temperature environments，the goodness-of-fit measure R² exceeds 0.9 for KNN，RF and GBDT. In terms of error，the KNN model has the smallest prediction error，with a root mean square error (RMSE) of 0.053 ℃. For the heat stress prediction model for police officers engaged in outdoor operations under high-temperature conditions，the predictive performance of RF，GBDT and KNN models is significantly better than that of other models.

In order to improve the durability of the construction machine and the working environment in deep buried tunnels，based on the theory of gas-solid two-phase flow，CO and dust were selected as the main objects of study，and a physical model of deep buried tunnels was established by Fluent software. The effects of different surrounding rock temperatures and the outlet speed of the wind pipe on the transport process of soot in deep tunnels were investigated through simulation. The results show that after blasting，CO is uniformly distributed in the throwing area. With the increase of ventilation time，the CO transport shows two modes of translation and diffusion. CO is discharged out of the tunnel in the form of a mass，and the CO transport speed at the tunnel wall is larger than that at the center of the tunnel. At the moment of blasting，a large amount of dust gathers near the working face，and with the increase of ventilation time，the dust is continuously discharged out of the tunnel. Among them，the temperature of the surrounding rock has a certain effect on the transportation of CO. The higher the temperature of the surrounding rock，the faster the transportation of CO. However，the effect of the surrounding rock temperature on the transportation of dust is relatively small. The outlet speed of the wind pipe has a greater impact on the transportation of CO and dust. The greater the outlet speed of the wind pipe，the faster the transportation of CO and dust. The field application should be combined with the actual conditions and economic budget to select the relevant equipment.

To quantitatively analyze the coupling relationship between risk factors of personal injury and death accidents in Chinese electricity production and identify key risk factors，these risk factors were categorized into four dimensions: human，machine，environment and management based on the 4M accident causation theory. Utilizing grounded theory，32 secondary risk factors were identified，and the coupling methods were categorized into three types: multi-factor，double-factor and single-factor risk coupling. Drawing from investigation reports of 196 personal injury and death accidents in Chinese electricity production from 2016 to 2022，the N-K model was used to measure the coupling level of risk factors，while DEMATEL method was used to pinpoint key risk factors. Finally，the countermeasures and suggestions for risk prevention and control were put forward. This study shows that the frequency of accidents is closely related to the coupling level of risk factors，and the coupling level of risk factors is proportional to the number of participating factors. The risk coupling value of the four types of risk factors is 0.162 8，indicating the highest coupling level. The risk value of human factors participating in the risk coupling is relatively high，followed by the risk values of management and environmental factors. Among the three-factor risk couplings，the human-machine-environment risk coupling has the highest risk value at 0.104 7. For the double-factor risk couplings，the human-environment risk coupling has the highest risk value at 0.05. Both risk couplings involve human risk factors. The failure to implement safety production investment and main responsibility，illegal operations and illegal command and decision-making are the key factors in the coupling effect of risk factors of personal injury accidents in electricity production，and priority should be given to prevention and control.