In order to seek the best way of laying median strip facilities，four typical forms of median strips in China were established through driving simulation，and trajectory and speed-related data were extracted. Based on the horizontal right of way in the road spatial right of way，the road utilisation rate was selected as an indicator to analyse the difference characteristics between the actual right of way and the nominal right of way. The results show that the overall range of fluctuations in driver trajectories at different height facilities is significantly different. Among the rightward offsets triggered by the shy away effect，the trajectory offset reaches the maximum in the reboundable traffic cylinders scenario R₅，followed by the traffic separate railings R₇ and raised pavement markers R₂，which are larger than that of the double yellow line scenario R₁ without the facility. While adding facilities triggers a shy away effect in drivers，a certain degree of lateral offset can improve road utilisation rate，up to 11.15%，which was found in R₅. At the same time，the installation of the facility inhibits the speed of drivers，achieving a two-way improvement in traffic safety and traffic design. Finally，the width of the median strip to satisfy the maximum nominal right-of-way utilisation was calculated to be 0.844 m and the facility's height to be 138.62 cm.

To address issues of correlation prediction indicators，outliers，and data imbalance in original data in rockburst prediction，a rockburst prediction method based on LLE-DBSCAN-SMOTE for data processing was proposed. Firstly，the maximum tangential stress of surrounding rock {\sigma }_{\mathrm{\theta }}，uniaxial compressive strength of rock {\sigma }_{\mathrm{c}}，uniaxial tensile strength of rock {\sigma }_{\mathrm{t}}，elastic strain energy index W_{\mathrm{e}\mathrm{t}}，brittle coefficient {\sigma }_{\mathrm{c}} / {\sigma }_{\mathrm{t}}，stress coefficient {\sigma }_{\mathrm{\theta }} / {\sigma }_{\mathrm{c}}，and stress concentration value β characterizing the stress gradient of surrounding rock were selected to construct a rockburst prediction indicator system. Secondly，the LLE algorithm was used for data dimensionality reduction to eliminate the cross-correlation effect between indicators，and the DBSCAN algorithm was introduced to remove outliers. Then，the SMOTE technology was introduced for data balancing. Finally，three types of rockburst prediction models were proposed using Decision Tree (DT)，Random Forest (RF)，and Gradient Boosting Decision Tree (GBDT) algorithms. The prediction accuracy of the data training models before and after processing was compared and analyzed. Moreover，engineering verification was performed through the measurement in the diversion tunnel of Jiangbian Hydropower Station. The results show that the prediction accuracy of the three types of algorithm models which reduce the prediction index from the 7 dimensions of the original data to the 4 dimensions and adopt the graded outlier processing is the highest among the similar models. The rockburst prediction of the Jiangbian Hydropower Station demonstrates that the proposed model significantly improves prediction accuracy compared to similar models using original rockburst data.

To improve construction workers' safety behavior and enhance the effectiveness of both implicit and explicit safety attitudes on this behavior，this study investigated the interaction between these two types of attitudes and their combined influence on safety behavior. First，an experiment was designed to measure the implicit safety attitudes of construction workers，and Implicit Association Test(IAT) was used to evaluate underlying attitudes. Then，the relationship between implicit and explicit safety attitudes was analyzed based on an explicit safety attitude scale. Finally，the study examined how the three components of both implicit and explicit safety attitudes—cognitive，emotional，and behavioral tendency—affected safety behavior. The results show that construction workers generally exhibit positive implicit safety attitude. However，the correlation between implicit and explicit safety attitudes is weak. Explicit safety attitude，particularly the overall，emotional，and behavioral components，has a significant positive effect on safety behavior，while the correlation between implicit attitude and safety behavior remains weak. When implicit and explicit safety attitudes are aligned，their correlation with and explanatory power for safety behavior increases.

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 reveal the influence of asymmetric load on gas seepage and extraction radius，a multi-physics coal and gas fluid-solid coupling model was proposed to analyze the gas seepage characteristics of coal seams. Matrix-adsorbed gas was used as the mass source in the proposed model introducing asymmetric loads into the boundary conditions. Furthermore，segmented drilling was used under asymmetric load conditions to optimize the gas extraction radius and improve extraction efficiency. The results indicated that greater stress compressed the cracks inside the concentrated stress zone，making it more difficult for gas to flow and to be extracted more challenging. The gas pressure in the concentrated stress zone decreased by approximately 2% less than that in the original stress zone，and the permeability decreased by about 9%. Asymmetric load had different degrees of influence on the diffusion and seepage processes. Within 180 days，the mass of diffused gas of the original stress area decreased by 19% and the seepage mass decreased by 20.5%，while these values in the concentrated stress zone decreased by 16.9% and 17.9%，respectively. Asymmetric loads had adverse effects on gas extraction，increasing extraction time under uniform load conditions. By adjusting the extraction radius under asymmetric load conditions，not only can the extraction efficiency be improved by approximately 3%，but it can also ensure that the extraction standards are met within 180 days，thereby effectively improving the overall performance of gas extraction.

To provide intelligent and systematic decision support for building safety management，building fire accidents data was collected and summarized. The knowledge graph of building fire accidents was developed to construct a knowledge graph database. Based on the dimensions of time，space，theme，and important entities，the implementation process of the intelligent question-answering system was innovatively presented. Moreover，the intelligent analysis of building fire risk was performed. The results showed that daytime and summer were high-risk periods for building fires. The frequency of building fire accidents in East China was significantly higher than that in other regions，and the fire risk of building fires was higher in electrical and warehouse areas. Reinforced concrete frame structures and factory buildings were more prone to building fires. Most ignition sources were combustible solids，and the main cause of fire accidents was illegal construction behavior.

In order to improve the recognition accuracy and detection speed of traffic participants by intelligent networked vehicles and traffic monitoring systems so that they can timely respond to the safety hazards in the mixed traffic environment in urban space，a mixed traffic participant detection model in urban space based on the improved YOLOv8n algorithm was proposed. Firstly，geometric transformation and pixel transformation enhancement strategies were employed in the data input stage to prevent overfitting and improve robustness，and generalization. Secondly，the SPD-Conv module was used to replace all original convolution layers of the YOLOv8n algorithm，which enhances the feature extraction capability for low-resolution small targets. Meanwhile，the CA module was added to the fusion structure of the neck network of the YOLOv8n algorithm to improve the recognition accuracy of key information with almost no additional computational overhead. Then，the boundary box loss function EIoU was used to replace the original loss function，enabling the model to achieve superior convergence speed and recognition stability. Finally，the ablation and comparison experiments were carried out with the public and self-built integrated traffic participant dataset，and the real-time detection experiment was carried out with the automatic driving experiment platform. The experimental results show that compared to the YOLOv8n model，the improved SEC-YOLO model has increased mAP and FPS by 3.2% and 7.9% respectively. The SEC-YOLO model outperforms mainstream models in terms of mAP and FPS as well. The average accuracy of real-scene detection on the automatic driving experimental platform is around 95%. The SEC-YOLO algorithm model achieves higher detection accuracy for urban traffic participants，with stronger robustness and real-time performance.

To solve the problems existing in the traditional NER methods in the domain of tunnel construction safety，such as fuzzy entity boundary，difficulty in small-sample learning，and insufficiently comprehensive extraction of feature information，an entity recognition method for tunnel construction accident text based on the BERT-BiLSTM-CRF model was proposed. Firstly，the BERT model was used to encode the tunnel construction accident text to obtain word vectors containing semantic features. Then，the word vectors output after the training of the BERT model were input into the BiLSTM model to further obtain the context feature of the tunnel construction accident text and conduct label probability prediction. Finally，by utilizing the constraints of the annotation rules of the CRF layer，the output result of the BiLSTM model was corrected，and the maximum probability sequence annotation result was obtained，so as to realize the intelligent classification of the labels of the tunnel construction accident texts. Comparative experiments were conducted between this model and other four commonly used traditional NER models on the tunnel construction safety accident corpus dataset. The results show that the recognition accuracy rate，recall rate and F₁ value of the BERT-BiLSTM-CRF model are 88%，89% and 88% respectively，and the entity recognition effect is better than other benchmark models. By using the established NER model to recognize the entities in the actual tunnel construction accident texts，its application effect in the domain of tunnel construction safety is verified.

In order to ensure the normal passage of vehicles and the safety of the existing tunnel support structure during the blasting through the highway，the evaluation method of engineering blasting effect based on extension-AHP model was proposed. Firstly，by means of investigation and analysis，the blasting effect rating standard and index system were established，and the model was applied to the evaluation of a water diversion project. Secondly，AHP was used to determine the weights of evaluation indexes，and the combined relevance degree of blasting rating was calculated. Finally，the results of the blasting effect rating were verified by acoustic detection test，blasting shock wave test and blasting seismic wave test. The study shows that the combined relevance degree {{{{{Q_j}_{(}}_X}_{-}}_M}_{)} is calculated by extension-AHP model. The blasting effect of the tunnel boring is Q_max=-0.017，and the evaluation grade is a good blasting effect. The surrounding rock loose circle of the tunnel is relatively small and evenly distributed. The influence range of the surrounding rock stability is about 0.5-0.6 m. The blasting energy does not cause the rock rupture zone to further extend the signs of the inward. The energy attenuation trend of blasting seismic waves is different under different wave frequencies. However，the attenuation rate is greater than that of low-frequency component energy in the overall performance of high-frequency component energy. In the same channel，with the increase of the distance between the blasting source and the measurement points，the overall vibration waveform becomes narrower. The main frequency increases first and then decreases，and the main frequency domain moves to the low-frequency direction. The overpressure peak attenuation characteristic of blasting shock wave meets P_S=αl^-γ. With the increase in the distance from the blasting source，blasting shock wave overpressure attenuation coefficient is an increasing trend. The measurement range belongs to the shock wave attenuation zone. The shock wave overpressure peak of the tunnel entrance and the construction outside tend to converge.