In order to optimize engineering fire-extinguishing technologies, numerical simulation was employed to investigate the effects of flow rate and pipe diameter at the foam inlet on the transport behavior of compressed air foam. In the simulation, the flow rate varied within the range of 1 000-2 400 L/min, and the pipe diameter ranged from DN80 to DN220. Characteristic parameters characterizing foam transport were extracted, including flow field, vorticity, viscosity, and pressure drop. The results show that the velocity of the foam fluid at the elbow forms three regions from the outside to the inside: low-speed, high-speed, and low-speed. However, the flow rate and pipe diameter changes have different degrees of influence on these three regions. Regarding the effect of inlet flow rate, there is a positive correlation between vorticity and inlet flow rate. Additionally, the low-speed region on the inner side of the elbow increases, the high-speed region is squeezed toward the outer wall, and the range of the low-speed region on the outer side decreases. Meanwhile, as the inlet flow rate of the foam fluid increases, the viscosity of the fluid gradually decreases, and the pressure drop first increases and then tends to stabilize. In terms of the influence of pipe diameter, compared with the inlet flow rate, the changes in velocity stratification and vortex near the elbow are relatively slight when the pipe diameter increases. Moreover, the pressure drop inside the pipe gradually decreases with the increase in pipe diameter. When the pipe diameter increases from DN80 to DN220, the local pressure drops decreases by approximately two-thirds.

To mitigate the occurrence of berthing accidents in fog, this study constructed a cloud model-based visual risk assessment framework for ship berthing operations under dense foggy conditions. Within this framework, a comprehensive risk assessment index system was first established, encompassing four dimensions: personnel involved in berthing, vessel characteristics, berth facilities, and environmental conditions. Subsequently, the combined AHP and DEMATEL method was employed to calculate the integrated influence weights of each assessment indicator. To address the uncertainty inherent in berthing risks, the cloud model was introduced to quantify the risks associated with foggy berthing operations. Finally, the proposed framework was applied to evaluate four typical berthing scenarios at the Waigaoqiao Container Terminal of Shanghai Port. The results demonstrate that the marine clearance (fore and aft distance) at the berth, the proficiency of pilots in foggy berthing operations, and the positioning effectiveness of berth leading marks or lights significantly influence the risk levels of ship berthing in fog. This risk assessment model accurately evaluates and visually represents berthing risks across different scenarios. It thus holds substantial importance for ensuring all-weather berthing operations and enhancing port cargo throughput capacity.

Runway excursion accidents (REAs) were represented as one of the most frequent threats to aviation safety. To enhance aviation safety management and optimize preventive strategies, it was essential to systematically understand the risk evolution mechanisms underlying REAs. This study develops a three-stage analytical framework-ARM, Weighted Complex Network Analysis, and causal inference based on Linear Non-Gaussian Acyclic Model with Independent Component Analysis (ICA-LiNGAM)-drawing upon 88 representative global REA accident reports (2008-2020) and Quick Access Recorder (QAR) data from 1 345 Boeing 737-700 flights at Dali Fengyi Airport. Firstly, ARM was utilized to identify frequent factor combinations and critical risk elements. Subsequently, a weighted complex network was established to quantify the coupling strength and network properties of these risk factors. Finally, the causal relationships among risk factors and the pathways of risk evolution were analyzed based on the ICA-LiNGAM causal inference method. The results indicate that go-around decision errors and excessive landing speed function as critical hub nodes dominating risk propagation. Environmental factors (e.g., crosswinds) and operational parameters (e.g., approach speed) exhibit bidirectional causal interactions, as evidenced by crosswinds significantly increasing landing speed (Effect value=0.201, p<0.001). This "human-machine-environment" cross-domain evolutiong triggers risk cascades—illustrated by high descent rates inducing runway directional instability (Effect value=19.713,p<0.05)-necessitating cross-domain collaborative interventions targeting critical nodes to disrupt propagation pathways.

In order to ensure the safe operation of gas pipeline networks and to enable the overall reliability assessment of pipeline systems, a three-level reliability evaluation model "component-segment-pipeline network" was proposed for natural gas pipeline networks. First, the state equations of steel pipe and weld units under different failure modes were established by analyzing the composition of the pipe network line system. Combined with the Monte Carlo algorithm, the failure probability under different failure modes was calculated. Based on fault tree analysis and the minimal cut set theory, a "component-segment" level reliability model was developed, enabling the hierarchical reliability evaluation from components to pipeline segments. Next, the GO method was applied to analyze the topological structure of the pipeline network, and a structural reliability model at "segment-pipeline networ" level was established. The structural reliability of the entire pipeline network was calculated and weak segments in terms of reliability were identified rapidly. Finally, a case study was conducted to verify the accuracy and applicability of the proposed model. The results show that the reliability of pipeline network systems is positively correlated with pipe material strength and wall thickness, and negatively correlated with the diameter-to-thickness ratio and internal pressure. Compared to high-grade pipelines represented by X80, networks composed of lower-grade pipelines exhibit higher sensitivity to external factors, with their reliability being more significantly affected by external disturbances.

In order to enhance the automatic recognition of safety hazards and improve safety management in construction scenarios, a multimodal large-model-based method for construction safety hazard recognition was proposed and its core component—the multimodal safety hazard recognition model, LLaVA(Large Language and Vision Assistant)-CS(Construction Site), was implemented. The system integrated images (construction site photos) with safety operating procedures (worker behavior descriptions), leveraging multimodal learning and deep learning technologies to perform real-time monitoring and analysis of construction sites. To support the system's effective operation, a multimodal dataset covering complex conditions such as varying lighting, occlusions, and multi-person scenarios was constructed, addressing the gaps in existing public datasets. Through prompt tuning of the LLaVA-1.5 model, the LLaVA-CS model effectively integrated visual and textual information, enhancing the accuracy and interpretability of safety hazard recognition. Experimental results show that this method achieves an accuracy of 0.722 2 in multiple real-world construction projects, generating detailed explanatory texts in real time to help managers quickly understand specific safety hazard contexts, thereby improving decision-making in safety management. This study innovatively applies multimodal large models to construction safety management systems, providing real-time, interpretable safety monitoring solutions and offering new technical support and optimization directions for construction safety management.

To improve the efficiency of safety inspections for existing civil buildings, this study develops a decision-support model for safety evaluation of existing civil buildings based on ML techniques. Building safety feature data were first collected from building evaluation reports. Then, a multi-dimensional indicator system integrating "design-evolution-status" features was established through correlation analysis and recursive feature elimination with cross-validation. Subsequently, five ML models were built and evaluated using performance metrics such as accuracy, precision, and recall. Furthermore, a decision-support platform for safety evaluation of existing civil buildings was developed and validated through a real engineering project to examine its practical operability. The results demonstrate that, compared to design features, evolution and status features more effectively reflect the actual safety conditions of civil buildings. In particular, building age, renovation or extension history, and concrete beam load capacity are identified as key features. Among the tested models, the Decision Tree algorithm shows the best performance in evaluating the safety of enclosure system, superstructure, foundation, and individual evaluation unit.

In order to investigate the decision-making mechanism for operational safety risks in bridges and mitigate major operational safety hazards, a three-stage theoretical framework comprising risk perception, risk cognition, and risk decision-making for bridge operational safety was established. H-OWA operator was employed to process expert evaluations, constructing an interaction matrix for decision factors. The meanings and structure of concept nodes within FCM were defined to elucidate the mapping relationships between factors. Monitoring data collected during train passages on a specific railway bridge on January 1, 2020, was utilized for conducting engineering application and validation analysis. The results show that the weight importance ranking of bridge structure acceleration, strain, and displacement is the highest among the iterative process importance ranking, indicating their critical roles in perceiving bridge structure health. Moreover, the implementation effect of reinforcement strategies for the substructure of the bridge is significantly higher than that of waterproofing and insulation and reinforcement of the superstructure under low, medium, and high-risk scenarios, effectively enhancing the overall stability and safety of the bridge.

This paper aims to understand the current situation of residents' flood EEC in Jiaozuo city, identify its influencing factors, and to improve their flood EEC. Firstly, a comprehensive index method was adopted to construct an index of residents' flood EEC from four dimensions: knowledge, skills, willingness and behavior. Then, based on literature analysis, ten key factors influencing residents' flood EEC were identified and analyzed. Finally, according to the conclusion, countermeasures and suggestions for improving residents' flood EEC were proposed. The results show that the flood EEC of urban residents in Jiaozuo city is at a medium to low level. The evacuation willingness score is the highest, the evacuation skills score is the lowest, and the evacuation knowledge and behavior scores are in the middle. From the perspective of influencing factors, male respondents, young people, those with higher education, higher income, more flood experience, large family members, more mobility-impaired family members, more children under 12 years old, those living in bungalows, and those living in flood-prone areas have higher flood EEC.

To address the issues of preference dependence and loss aversion of decision makers under risk conditions in the evaluation of urban flood disaster emergency management efficiency, an improved grey group decision-making model integrating the risk preference of decision-makers was constructed. Firstly, a decision-maker weight determination method based on group consensus and difference was proposed, which overcame the limitation of traditional method that ignored differences of decision-maker's professional background, experience and personal preference when integrating individual information. Then, the TODIM method was proposed to reduce the problem of loss overestimation faced by decision makers under risk when calculating dominance. Finally, the model was applied to the efficiency evaluation of urban flood disaster emergency management schemes, which verified its effectiveness. The results show that the scheme of improving the emergency response and rescue system is superior to both the early warning and monitoring system construction scheme and information technology support scheme, when comprehensively considering emergency rescue efficiency, disaster prevention and mitigation effect, resource utilization efficiency and emergency management cost. Moreover, this optimal emergency scheme remains robust in 81.82% of the scenarios.

In order to enhance safety management within urban rail transit stations, a passenger flow detection and early warning model based on the YOLO11-DeepSort algorithm was proposed. First, a pedestrian head dataset was established and put into YOLO11 to train model parameters for extracting passenger flow density. Then, DeepSort was used for dynamic tracking of passengers' walking paths to extract parameters of passenger flow and walking speed. According to the parameters of passenger flow status at station bottlenecks such as transfer passageways, stairs and platforms, FCM algorithm was used to divide the passenger flow status into 6 subcategories, and 4 early warning levels were divided corresponding to the passenger flow status. Finally, taking Guanyinqiao Station of Chongqing Rail Transit as an example, the first 4 days of real-scene station videos were used to identify passenger flow and extract parameters, and the FCM model was trained based on those parameters to get clustering centers and thresholds. By using the model after training, the 5^th day's real-scene video was classified, and the classification results of full-day, peak and off-peak periods were compared. The results show that the platforms in the station have the longest congestion time and the greatest risk, with the peak-hour congestion duration accounting for 70% and the Level 1 early warning duration accounting for 15%. Stairs in the station have the second longest congestion time and risk, with the peak-hour congestion duration accounting for 46% and the Level 1 early warning duration accounting for 10%. Passageways in the station have shorter congestion time and lower risk, with the peak-hour congestion duration accounting for 41% and the Level 1 early warning duration accounting for 5%.