Search Results (51)

To reveal the influencing mechanism of safety resilience in prefabricated building systems (PBS), identify key risk nodes, and support targeted resilience enhancement, this study develops an improved ISM–BN analytical model. Based on 136 domestic safety accident cases involving prefabricated buildings (PB) from 2016 to 2025, and combined with bibliometric analysis, 13 causal factors were identified and an indicator system was established. Grey Relational Analysis (GRA) was introduced to improve the traditional Interpretive Structural Modeling (ISM) method, through which the causal factors were divided into four hierarchical levels and the hierarchical relationships among the factors and levels were clarified. Subsequently, the hierarchical structure derived from the improved ISM was mapped into a Bayesian Network (BN), and parameter learning was conducted using accident data. Through backward diagnosis and sensitivity analysis, five key risk nodes and two critical transmission paths were identified, based on which targeted improvement strategies were proposed. The results can provide methodological support and decision-making references for key risk control and resilience enhancement in PBS. Full article

Accident diagnosis in nuclear power plants (NPPs) should provide mechanism-aligned evidence that can be reviewed by operators and safety engineers, rather than only a high-confidence accident label. Existing data-driven methods achieve strong classification performance but often express explanations as attention maps, anomalous nodes, prototypes, or causal links separately, making it difficult to obtain a unified diagnostic evidence chain. To address this limitation, we propose a Concept-Constrained Physical Graph (CCPG) framework that formulates accident diagnosis as structured evidence generation. CCPG groups multivariate transient signals into operator-readable physical nodes, extracts node-wise temporal features, and propagates them over a mechanism-guided graph. It then couples a concept bottleneck with implicit latent features, prototype learning, and edge/stage supervision to predict the accident class and a reviewable evidence package. On the evaluated NPPAD five-class simulated benchmark, CCPG achieved saturated clean-set classification (1.000 accuracy) and high paired-challenge accuracy (0.996) while providing concept, affected-node, edge-template, stage-order, and prototype evidence. Additional analyses, including Transformer baselines, feature-restricted and early-window stress protocols, calibration, statistical testing, open-set detection, and scalability profiling, further characterize the robustness, reliability, and applicability of the proposed framework. Full article

Trustworthy deployment of artificial intelligence in safety-critical systems requires accurate diagnosis of anticipated scenarios and reliable rejection of out-of-distribution (OOD) inputs that fall outside the modeled operational scope. Existing data-driven diagnostic models typically assume that test inputs are drawn from the training distribution or rely on heuristically tuned thresholds that lack enforceable safety guarantees. This article presents SCOPE (Safety-Calibrated Out-of-distribution Prediction via Contrastive Embeddings), a framework integrating supervised contrastive learning with split-conformal prediction to provide statistically grounded OOD rejection with finite-sample false-alarm control. SCOPE employs a causal residual convolutional encoder to map multivariate sensor streams into a hyperspherical embedding space with a compact, class-specific structure. A k-nearest-neighbor density nonconformity score, computed in the encoder embedding space, flags transients that occupy low-density regions relative to known accident manifolds; an ablation shows that this density score outperforms prototype distance, entropy, and conservative maximum fusion as well as a panel of standard OOD baselines (MSP, ODIN, energy, Mahalanobis, OpenMax, MC-dropout, and a reconstruction autoencoder). To support temporally evolving trajectories, SCOPE aggregates window-level scores under a monotone decision policy and performs trajectory-level conformal calibration, yielding distribution-free guarantees that bound the probability of falsely rejecting a known accident run. SCOPE is evaluated on the Nuclear Power Plant Accident Data (NPPAD) benchmark using high-openness splits that withhold entire accident families as unknowns, and all metrics are reported as mean ± standard deviation across multiple random seeds. Results demonstrate strong diagnostic accuracy on accepted trajectories, conservative false-alarm rates satisfying user-specified safety constraints across multiple operating points, and timely rejection of unseen accident mechanisms, making SCOPE suitable for deployment in safety-critical monitoring applications. Full article

To address the limitations of incomplete factor identification, insufficient cross-level coupling quantification, and inadequate causal path verification in traditional human factor analysis of road transportation of dangerous goods (RTDG) accidents, this study developed an integrated HFACS-Apriori-SEM analytical framework that enables full-process analysis from factor identification to causal mechanism exploration and hierarchical path validation. A five-level industry-specific Human Factors Analysis and Classification System (HFACS) framework with 85 causal indicators was established, and standardized coding was conducted for 58 fatal RTDG accidents in China from 2012 to 2022. Twelve core strong association rules were generated using the Apriori algorithm. Among these, the co-occurrence chain “organizational process failure → inadequate supervision → insufficient personnel readiness → routine violations” had the highest support of 0.621. Structural equation modelling (SEM) provided empirical support for a significant hierarchical chain transmission effect of the accident causation. The findings showed that preconditions for unsafe acts exerted the largest indirect effect on accident severity (total effect = 0.69, p < 0.001). Furthermore, unsafe acts were the only direct influencing factor (total effect = 0.85, p < 0.001). In addition, violations accounted for a significantly higher proportion of unsafe acts than errors. This study provides strong empirical evidence that catastrophic RTDG accidents stem from the chain failure of multi-level system defenses, offering a quantitative and targeted decision basis for hierarchical accident prevention and control in the RTDG industry. Full article

Railway cable-stayed bridge construction is characterized by high complexity and substantial safety risk. Deficiencies in safety control may result in serious accidents (e.g., collapse and falls), causing significant casualties and economic losses; therefore, clarifying risk interactions and accident-causing mechanisms is essential. This study proposes a fuzzy DEMATEL–ISM approach in which fuzzy sets capture uncertainty in experts’ linguistic assessments. DEMATEL quantifies influence strengths and causal relationships among factors, and ISM constructs a multi-level hierarchy to explain accident causation. Twenty safety influencing factors are identified and grouped into five categories: management, human, material and equipment, construction technology, and environmental conditions. The obtained accident-causing mechanism comprises seven hierarchical levels: L₁: collapse and fall accidents, L₂: direct factors, L₃–L₅: indirect factors, and L₆–L₇: root factors. This mechanism is a chain of events that leads to an accident, with the nodes improper prestressing, structural deformation and differential settlement. These key nodes can be avoided by reinforcing safety management system implementation, daily supervision and inspection, and education and training on the subject of safety to ensure the safety of railway cable-stayed bridge construction. Full article

The analysis of dynamic causal mechanisms underlying shipboard fires and explosions is often restricted by the unstructured and fragmented nature of accident investigation reports. This study proposes a framework integrating ontology-driven information extraction with Dynamic Bayesian Networks (DBNs) to model temporal accident evolution. An ontology comprising 41 nodes was constructed through a structured expert elicitation process to formalize the domain knowledge. To process 198 bilingual accident reports, an extraction pipeline was deployed, incorporating XLM-RoBERTa, BiLSTM-CRF, and an entity-marker relation classifier. Large language model (LLM)-directed weak supervision, constrained by token-level information entropy filtering, was employed to expand the training corpus, necessitating only 2.5% manual verification. The extracted semantic dependencies were utilized to initialize a three-slice DBN (precursor, initial fire, and escalation/explosion). The network structure was jointly optimized through ontology constraints (112 forbidden and 4 mandatory edges), the Hill-Climbing algorithm, and BDeu scoring. The proposed DBN achieved an AUC of 0.759 ± 0.086 and a Brier Score of 0.192 ± 0.021 (1000 bootstrap iterations), demonstrating superior predictive performance over traditional interpretable models (Static BN, HMM, ETA) with large effect sizes (Cohen’s d > 1.0), while maintaining competitive accuracy and enhanced causal interpretability relative to XGBoost. This framework offers a scalable, data-driven methodology for dynamic probabilistic risk assessment in maritime safety. Full article

A coal mine gas explosion is a systematic failure caused by the interaction of multiple factors. In previous studies, most research determined the key causes based on practical experience or a single static indicator. This study puts forward a comprehensive method that integrates complex network theory and a genetic algorithm. By analyzing the explosion mechanism, a network model with 43 causal factors as nodes and their relationships as edges was established, thus capturing the overall structure of the accident system. Subsequently, the genetic algorithm was employed to optimize the identification of key nodes in the network. At present, most of the research on accident risk assessment relies on static topological analysis, failing to take into account the synergistic effects resulting from the simultaneous removal of multiple nodes, and is prone to getting stuck in local optimal solutions. The purpose of this study is to be able to search for the most influential node set and reduce the reliance on static indicators. The results show that both random attacks and deliberate attacks can reduce network efficiency. Meanwhile, when attacking the key cause combinations identified through searching, the network efficiency drops most rapidly. This indicates that the network is more vulnerable in more targeted attacks. This method encourages us to transition from a single-dimensional risk assessment to a comprehensive and multi-dimensional analysis framework. Full article

Accident reports provide a detailed account of environmental causes, unsafe human behaviors, and subsequent chain reactions. These records serve as essential resources for analyzing accident mechanisms and exploring potential risk patterns within production safety processes. Currently, Graph based Retrieval-Augmented Generation (RAG), which integrates Large Language Models (LLMs) with Knowledge Graphs (KGs), has emerged as a leading approach for complex causal question answering over extensive unstructured accident documentation. However, the application of this technology in the production safety domain still encounters two primary challenges. First, knowledge graph construction using a single granularity fails to capture fine-grained case details and macro-level standard systems. Second, traditional one-step retrieval paradigms lack the capacity to track deep causal chains or interpret the complex logic of multi-factor coupling. To address these limitations, we propose CausalAgent, a hierarchical graph-enhanced multi-agent framework for causal question answering in production safety accident reports. This framework innovatively combines a Hierarchical Causal Graph (HC-Graph) and a Multi-Agent Collaborative Reasoning (MACR) mechanism. Specifically, the HC-Graph employs a two-layer architecture that links a fine-grained instance layer with a national standard causation layer to resolve conflicts in semantic granularity. The MACR mechanism converts complex natural language queries into executable structured queries and logic verification steps through the sequential cooperation of four specialized agents, namely the Graph Parsing Agent, the Problem Analysis Agent, the Query Generation Agent, and the Reasoning Insight Agent. CausalAgent enables in-depth mining of accident causation mechanisms and provides scientific, robust and interpretable intelligent support for data-driven risk assessment and emergency decision-making. Experiments on real-world accident datasets demonstrate that CausalAgent achieves a $100.0\%$ query execution rate and an $87.3\%$ reasoning accuracy, outperforming the SOTA baseline by $45.2\%$ in terms of absolute accuracy. Full article

The implementation of the national emergency industry demonstration bases’ policies is a new way to achieve safety production governance and a key factor in improving the effectiveness of national safety production governance. This study regards China’s national emergency industry demonstration bases’ policies as a quasi-natural experiment. Based on panel data from 31 provinces in China from 2010 to 2022, a multi-period difference in differences (DID) model is conducted to systematically evaluate the impact and mechanism of this policy on China’s safety production governance. The results show that this policy significantly reduced the death rate of safety production accidents with a GDP of 100 million yuan and has a significant governance improvement effect. Further analysis of the mediating effect shows that policies mainly exert governance effects by increasing public safety financial investment and promoting innovation output. The heterogeneity analysis results indicate that policy effects are more significant in regions with weaker energy-resource industrial bases and lower levels of digital development, suggesting that the marginal governance benefits of policies are mainly concentrated in areas with relatively weak supporting conditions for safety governance. This study makes three primary contributions to the literature. Theoretically, it expands the safety governance paradigm by shifting the focus from traditional administrative “command and control” regulations to market-driven industrial agglomeration. Methodologically, by utilizing a multi-period DID model, it overcomes endogeneity issues prevalent in prior correlation-based studies to rigorously identify causal effects. Empirically, it opens the “black box” of policy transmission by validating dual pathways—fiscal resource allocation and technological innovation—while highlighting a critical “filling the gap” marginal utility effect in resource-constrained regions. This study empirically reveals the mechanism and context-dependent characteristics of industrial policies in safety governance, providing empirical evidence for understanding the inherent logic between industrial policies, public safety governance, and regional sustainable development. It offers practical insights for optimizing the precise implementation and resource allocation of emergency industrial policies to foster socially sustainable and resilient industrial growth. Full article

This paper presents a novel system-oriented counterfactual deep learning framework, termed Hybrid Prediction–Intervention Neural Architecture (HPINA) for intelligent traffic accident risk prediction and proactive safety intervention in smart transportation systems. Unlike conventional data-driven models that rely solely on observational correlations, the proposed system integrates multi-domain data fusion, temporal deep representation learning, a continuous spatio-temporal risk field, and a latent-space counterfactual reasoning module within a unified decision-support architecture. The framework enables accurate prediction of traffic accident risk and simulation of “what-if” intervention scenarios to support real-time safety optimization in intelligent transportation environments. By leveraging heterogeneous inputs, including traffic dynamics, environmental conditions, road attributes, and temporal patterns, the system constructs a high-dimensional representation that captures complex nonlinear dependencies and evolving risk propagation across the network. A key innovation lies in the integration of a causal intervention mechanism and policy-guided decision layer, which jointly quantify intervention impact and identify optimal strategies for minimizing risk. The experimental results demonstrate that HPINA achieves a Test F1-score of 0.958 and an AUC of 0.989, outperforming strong baselines by up to 5.0% and 3.4%, while achieving a relative risk reduction of 0.091 and improved convergence stability with a validation loss of 0.042. These findings highlight the effectiveness of the proposed framework as an intelligent, scalable, and deployable system for real-world traffic safety management and smart city applications. Full article

Maritime Autonomous Surface Ships (MASS) introduce new safety challenges associated with complex cyber–physical systems, distributed control architectures, and remote supervisory operation. Traditional maritime risk assessment approaches primarily focus on component failures and historical accident data and may therefore be insufficient for capturing interaction-driven hazards arising in autonomous vessel systems. This study develops a parallel and architecturally synchronized risk assessment framework integrating System-Theoretic Process Analysis (STPA) and Fault Tree Analysis (FTA) for the safety assessment of MASS. Within the proposed framework, both analyses evolve concurrently within a shared system architecture, enabling explicit traceability between hazards, unsafe control actions, causal scenarios, failure events, and accident propagation pathways. The framework is demonstrated through a case study of a Degree of Autonomy 3 short-sea freight vessel operating in a high-density North Sea traffic environment. The integrated analysis identifies dominant accident pathways related to perception degradation, communication disturbance, authority coordination conflicts, maneuver execution deviations, and incorrect collision-risk assessment. The results illustrate how the framework supports structured safety assessment of MASS while preserving traceability between systemic control deficiencies and accident propagation mechanisms. Full article

To unveil the underlying risk structures in complex industrial systems, this paper proposes a hybrid analytical framework that integrates BERTopic modeling, a large language model (LLM), and social network analysis (SNA). This framework aims to extract systemic safety intelligence from unstructured accident reports. It first employs BERTopic to identify latent causal topics based on 745 Chinese accident investigation reports and utilizes DeepSeek-V3.1 (LLM) for semantic refinement and causal mapping of these topics. Subsequently, a semantic network of causal keywords based on positive pointwise mutual information (PPMI) is constructed, and its topological structure is analyzed using SNA methods. The study identifies and analyzes five major risk communities: confined spaces, fire, mining, construction, and road traffic. It reveals that accident causation exhibits the small-world characteristics of multi-factor coupling and non-linearity, with core risk nodes concentrated in systemic inducements such as organizational management and compliance deficiencies. The results demonstrate that this framework effectively identifies the latent systemic risk patterns embedded within the texts, providing methodological support for developing sustainable safety management mechanisms based on design for safety. Full article

To identify and evaluate the failure issues in the livestock manure-to-biomethane supply chain, this study employs a Bayesian network approach with three inference analysis methods: diagnostic analysis, sensitivity analysis, and maximum causal chain inference. First, the main hazard categories affecting the failure of the supply chain are identified, establishing risk indicators for feedstock collection, pretreatment, anaerobic digestion, purification and upgrading, transportation, and biomethane end-use. Then, the half-interval method and possibility superiority comparison are used to calculate and rank the severity of related accidents, obtaining the severity ranking of secondary indicators as well as the severity ranking of work items and risk items. Finally, Bayesian forward inference is applied to investigate the failure probability of the supply chain, combined with backward inference to identify the risk factors most likely to cause supply chain failures and trace the formation of failure hazards. The Bayesian sensitivity analysis method is ultimately applied to determine the key hazards affecting supply chain failures and the correlations between accident hazards, followed by validation. The results show that the failure probability of the supply chain through causal inference is approximately 54.76%, indicating relatively high failure risk. The three factors with the highest posterior probabilities are mechanical stirring failure C₃ (88.11%), corrosion-induced ammonia leakage poisoning D6, and equipment explosion caused by excessive pressure due to overheating during dehumidification heating D₉, which are the hazards most likely to cause failures in the supply chain. Improper operations and the toxicity of related chemicals are key hazards leading to supply chain failures, with the correlation between accident hazards presented as a hazard chain by integrating severity and accident probability, and the key risk points in the supply chain are identified. Full article

Unsafe behaviors among construction workers remain a leading cause of accidents in the construction industry. Previous studies have primarily relied on structural equation modeling and causal inference approaches to investigate the determinants of workers’ safety behavior. However, these methods are often limited in their ability to address confounding bias inherent in observational data and tend to focus on isolated effects of individual variables, thereby overlooking the complex interactions between organizational and individual factors. To overcome these limitations, this study applies the Categorical Boosting (CatBoost) algorithm to examine the joint organizational and individual mechanisms underlying construction workers’ safety behavior. CatBoost is particularly suitable for small- to medium-sized datasets and is capable of automatically capturing complex, nonlinear relationships among variables. Leveraging the SHAP interpretability framework, both main-effect and interaction analyses are conducted to systematically identify the most influential determinants. The results demonstrate that CatBoost outperforms eXtreme Gradient Boosting (XGBoost) and Random Forest (RF) models in predicting safety-related outcomes. Prosociality (PSO) is identified as the most influential predictor, followed by personal proactivity (PAC). Interaction analyses further reveal that organizational attributes—such as prosociality, loyalty, and mutual assistance—play a critical role in cultivating a safety-oriented organizational climate, while an optimistic personal attitude further enhances safety performance on construction sites. Overall, these findings provide meaningful theoretical insights and practical implications for improving safety management in the construction sector. Full article

The construction industry faces severe safety challenges with over 80% of accidents stemming from unsafe behaviors, yet traditional management overlooks the role of individual differences, and existing research fails to address the specific psychological mechanisms operative in this high-risk, dynamic environment. To effectively curtail unsafe behaviors in such high-risk environments, this study aims to reveal the underlying mechanisms through which personality traits influence unsafe behaviors. Grounded in causal chain theory, the theory of planned behavior, and trait activation theory, this study constructs a hypothetical model of personality traits and unsafe behaviors, with fluke mentality serving as a mediating variable and safety climate as a moderating variable. A comprehensive approach combining questionnaire surveys, confirmatory factor analysis, correlation tests, and linear regression was employed to test the hypotheses. The results indicate that neuroticism, openness, and extraversion have significant positive effects on unsafe behaviors, while conscientiousness has a significant negative effect; agreeableness shows no significant influence. Fluke mentality plays a partial mediating role between personality traits and unsafe behaviors, while safety climate plays a negative moderating role. By clarifying the cognitive pathways of individual differences, this study enriches the theoretical framework of unsafe behavior research. The findings provide a theoretical basis for construction enterprises to optimize safety management from the perspective of individual differences, offering practical pathways to promote high-quality development in the construction industry. Full article