To investigate the failure issues of overdue in-service pressure pipelines, this paper proposed a method based on integration of BOA and SAM for analyzing the failure probability of such pipelines, building upon the FTA framework. First, the causes of failure in overdue in-service pressure pipelines were analyzed, and a fault tree for failure was established. Then, experts were invited to evaluate the basic events of fault tree, and their opinions were aggregated using SAM method and an improved BOA-SAM approach. The accuracy of aggregation was assessed by comparing the uncertainty values of two methods (the lower the uncertainty value, the more precise the aggregation result). Finally, the failure probability of the top event was calculated, and an importance analysis was conducted. The results show that the uncertainty values of the aggregation results obtained using the improved BOA-SAM are significantly lower than those of SAM method. Notably, for the basic event X₆, the uncertainty value decreased by 7.6%. The failure probability of the top event is calculated to be 0.117 641. Stress corrosion cracking, third-party damage, weld cracking and excessive H₂S content in the medium are identified as the key factors contributing to the occurrence of accidents. This study provides a scientific and logical approach for risk assessment and failure analysis of overdue in-service pressure pipelines.

In order to improve the systematicity and objectivity of safety risk identification in civil aviation ATC operations and to explore the causes and interaction relationships of ATC unsafe incidents, a causal analysis method for ATC unsafe incidents was proposed, integrating TM, HFACS, and FRAM. Firstly, TM technology was employed to perform text processing on unsafe incident reports, and high-frequency keywords related to ATC operation risks were extracted using the Term Frequency-Inverse Document Frequency (TF-IDF) method. Secondly, semantic attribution was conducted on these keywords, and an improved multi-level HFACS framework for ATC operations was constructed based on the characteristics of ATC work, and the corresponding risk factors were determined. Then, the interdependent relationships between various functional modules in the ATC operation system were analyzed by introducing FRAM to reveal the potential impact of interactions within the complex ATC system on incident occurrence. Finally, two cases of ATC unsafe incidents were investigated to analyze their risk causes. The results show that the proposed method effectively mines the interrelationships between multi-source causes and identifies multi-level risk causal factors, which verifies the effectiveness and scientific rigor of the method in multi-source data processing and complex causal structure mining.

To address the limitations of traditional models in mining multi-dimensional temporal features and handling class imbalance, A multi-task driving risk prediction model based on Class Balance loss (CB)- Asymmetric Loss (ASL)-TimesNet and is proposed. A sliding time window was employed to extract multi-dimensional temporal features, improving the objectivity and granularity of risk level labeling. Based on these features, four types of labels were constructed: forward collision risk, rear collision risk, lateral collision risk, and steering intention. In terms of model design, the TimesNet architecture was incorporated to effectively capture periodic variations and dynamic evolution in multi-dimensional temporal features, thereby enhancing the modeling capability of temporal information in complex driving scenarios. Meanwhile, a hybrid loss function integrating CB and ASL was devised to improve prediction performance under class-imbalanced conditions. Experimental results demonstrate that the proposed CB-ASL-TimesNet model achieves an average accuracy of 0.908 6 in driving risk and steering intention prediction tasks. Compared with the traditional machine learning model CatBoost, the proposed approach yields a 16% improvement, and it outperforms the mainstream time-series model Gate Recurrent Unit(GRU) by 5.8%, verifying the significant effectiveness of the proposed model in enhancing prediction performance.

In order to effectively identify the unsafe behaviors of construction workers in high-altitude environments, a recognition model based on MCT-GCN was proposed. Firstly, a data preprocessing module was designed to convert video surveillance data into three-dimensional skeleton data. Secondly, a tri-component dynamic adjacency graph convolution was constructed. It integrated learnable structural prior topology, channel correlation adaptive topology, and relative position encoding topology to dynamically build adjacency matrices adapted to different actions. Furthermore, a multi-scale separable temporal convolution was proposed to decompose standard convolutions into deep temporal convolution and point-by-point convolution, thereby independently modeling the temporal characteristics and spatial distribution characteristics of construction workers' actions. Finally, experimental verification and comparative analysis were conducted on both public datasets and a self-built dataset of workers' unsafe behaviors. The results demonstrate that the proposed model outperforms existing methods in terms of recognition accuracy and cross-scene generalization. On the self-built dataset, the model achieved a peak recognition accuracy of 95.8%, significantly enhancing the detection of workers' unsafe behaviors in complex and dynamic construction environments, and making a significant contribution to the development of intelligent monitoring and control in the construction industry.

To address the unsustainable problems arising from significant demand gaps, inadequate support, and delayed supply of emergency resources in major disasters, and to tackle the limitation of traditional evaluation indicators that overlook social and environmental sustainability, this study was conducted. Firstly, based on the triple bottom line theory of sustainable development and literature mining method, a comprehensive evaluation index system was constructed from three dimensions: economic support capacity, social support capacity, and environmental support capacity for emergency resource support. Then, the original indicators were revised through expert evaluation to determine the final selected indicators. Finally, the grey relational analysis method was applied to construct an evaluation model for emergency resource support capacity in major disasters oriented towards sustainable development. Taking Henan Province as an example, the model was applied to verify its scientificity and effectiveness. The research shows that, when considering various factors such as economy, society, and environment, the correlation degree of social support is 0.933 90, having the strongest impact on emergency resource support capacity. In addition, indicators such as fixed assets investment in the logistics industry, the number of social organization units, the number of hospital beds in medical and health institutions, and regional GDP rank among the top, making them key influencing factors driving emergency resource support in Henan Province.

The rapid growth of passenger volume of urban rail transit and the urgent need for intelligent operation have made accurate section passenger flow prediction a key technical challenge to improve the level of dynamic scheduling and safety control. To this end, this paper innovatively integrated the physical information constraint mechanism, data-driven method and SGA, and proposed a new deep learning framework. Firstly, a physical residual term was designed and embedded into memory cells as a regulation signal, forcing the model to learn the physical laws of passenger flow while retaining the temporal characteristics of passenger flow. Secondly, a dual-objective fitness function based on physical loss and data loss was innovatively proposed to achieve further optimization of model performance while establishing a constraint mechanism. Finally, SGA was used to balance the differentiated and synergistic effects of hyperparameters in the model. Experimental results show that the constructed model exhibits good predictive performance on both the training set and the validation set. The improved fitness function can narrow the error range of the model prediction results. In the two-stage ablation experiment, the mean square error range of the proposed deep learning framework is reduced by 71.03% compared with the long short-term memory model, which verifies the synergistic enhancement of the model prediction ability by the simultaneous introduction of physical constraint mechanism and intelligent optimization algorithm.

In order to accurately predict the noise distribution of offshore platforms to prevent occupational noise hazards, a joint simulation method combining the sound ray method and SEA method was proposed to solve the problem that the traditional single noise prediction method was difficult to take into account the structural sound and air sound propagation characteristics and could not describe the large-scale spatial sound field distribution in detail. In this method, SEA was used to simulate the acoustic propagation characteristics of the structure, and the acoustic propagation law of the air in the large space of the deck was simulated by the sound ray method. Taking an offshore central platform as the research object, a joint simulation model was constructed based on the measured vibration and noise source intensity data. The sound field distribution of each deck was simulated and calculated, and the effectiveness of the method was verified by a spatial grid noise test. The results show that the contribution of air acoustic energy to the total noise energy of each deck of the offshore platform is more than 50 %, which is higher than that of structural acoustic energy. Obstacles such as large-scale equipment, production area rooms and firewalls have a significant effect on noise shielding, and air sound propagation has a more prominent impact on the spatial sound field distribution of the deck, the joint simulation results are in good agreement with the measured results, and the overall average difference between the prediction and the test data is within 2 dB, which can realize the high-precision prediction of the spatial sound field distribution of the offshore platform deck.

In order to scientifically identify the key disaster-causing factors and key disaster-causing paths in the urban rail transit system's flood disaster chain, and enhance the resilience of these systems, a study was conducted based on news report text data related to the urban rail transit flood disasters from 2010 to 2024. Natural language processing technology and a rule-based template matching method were used to extract causal event pairs. The t-Distributed Stochastic Neighbor Embedding (t-SNE) method was used to generalize the extracted event pairs, and the abstract event evolutionary graph of rainstorm disaster in the urban rail transit system was constructed, which was visualized as a disaster chain evolution network combined with Gephi. The complex network theory was introduced to quantitatively analyze the rainstorm disaster chain evolution network of the urban rail transit system, and identify the key disaster factors and key disaster paths in the disaster chain. The results show that the four types of disaster events, rainstorm, station water, subway shutdown and equipment failure, are of high importance and are the key disaster factors in the rainstorm disaster chain evolution network of the urban rail transit system. The three evolutionary paths of "road water → station water", "subway shutdown → economic loss" and "road water → rainwater backpouring" have high vulnerability, which are the key disaster paths in the rainstorm disaster chain evolution network of the urban rail transit system. The identified key disaster-causing factors and key disaster-causing paths can provide decision support for disaster prevention and mitigation.

To enhance the efficiency of emergency supplies distribution after an earthquake, an emergency vehicle routing optimization problem was investigated under limited transportation capacity and the need to simultaneously deliver multiple categories of emergency supplies, while road conditions and demand urgency were jointly considered. Firstly, vehicle travel speeds were corrected based on road damage rates, and a demand urgency evaluation method is established by incorporating key characteristics of earthquake disasters. Then, a post-earthquake emergency supplies distribution optimization model was formulated to minimize the total delivery time and the total urgency-weighted demand cost, involving multiple depots, multiple affected sites, multiple types of emergency supplies, and heterogeneous vehicle fleets. Next, a HEA integrating the fast non-dominated sorting genetic algorithm (NSGA-II) with variable neighborhood search (VNS), is then designed to solve the model. Finally, a case study based on the 2008 Wenchuan (5.12) earthquake was constructed, and simulation experiments were conducted to validate the effectiveness and feasibility of the proposed model and algorithm. The results show that HEA outperforms three benchmark multi-objective optimization algorithms in both solution-set convergence and overall solution quality, and can provide emergency decision-makers with a diverse set of trade-off solutions within a short computation time (average 115.0 s).

In order to promote the implementation of the responsibilities of participating entities in the prevention and handling of mine safety accidents, improve the efficiency of safety governance and reduce the risk of mine safety accidents, the assumption of an enterprises' illegal liability was analyzed. The authority and responsibility system of regulatory authorities was sort out. The guarantee mechanisms for the professionalism and objectivity of work safety social service institutions were explored. Drawing on the provisions and practical experience of foreign jurisdictions regarding the primary responsibilities for work safety, and putting forward legal suggestions for improving the prevention and handling of mine safety accidents in China. The findings show that the addition of punitive damages and differentiated penalties for repeated illegal acts are important factors for enterprises to consider the costs and liabilities of illegal acts, which can raise enterprises' awareness of the importance of wore safety and firmly establish their sense of responsibility for work safety. Allocating the responsibilities of regulatory authorities by stages and constructing a regulatory model that attaches equal importance to inspections and unannounced investigations can straighten out the predicament of unclear division of powers and responsibilities of regulatory authorities and improve regulatory efficiency. Refining the access threshold for work safety social service institutions, establishing an information disclosure platform, and clarifying the provisions on accountability and the practice withdrawal system can ensure the professionalism, transparency and objectivity of the services provided by work safety social service institutions.