广州航海学院学报

基于中国海事局144份船舶自沉事故调查报告，通过事故原因频次统计筛选25项基本事件，运用风险矩阵法量化风险等级，直接识别关键因素。结合致因链分析并专家评审，构建事故树模型，揭示事故演化逻辑，并转化为贝叶斯网络模型，通过双模型联合，推演关键致因路径。研究表明：船舶自沉的核心直接因素包括恶劣天气海况船体锈蚀/破损，冒险航行、未履行报告义务及非法采砂/运输活动等人为因素尤为突出。具体致因链如下：物理链方面，恶劣天气/船体老化导致结构破损进水，叠加应急响应失效引发自沉；人为链方面，非法作业驱动冒险航行，并叠加安全管理缺失加剧险情失控，最终导致致沉。事故树与贝叶斯网络耦合揭示了多因素动态作用机制，为复杂海事事故致因研究提供了新视角。研究结果可为船舶安全设计优化、高风险作业监管及航行制度完善提供依据。

Based on 144 investigation reports of ship self-sinking accidents by China Maritime Safety Administration, this study screened 25 basic events through accident cause frequency statistics, and directly identified key factors by quantifying risk level using risk matrix method. Combined with causal chain analysis and expert review, the accident tree model was constructed to reveal the accident evolution logic and transformed into a Bayesian network model to jointly deduce the key causal paths through the dual model. The core finding is that the core direct factors of ship self-sinking are severe weather and sea condition and hull corrosion/damage. The human factors, such as risky navigation, failure to fulfil the reporting obligation and illegal sand mining/transportation activities, are particularly prominent. The specific causal chains are as follows: physical chain-severe weather/hull deterioration leading to structural damage and water ingress, superimposed on the failure of emergency response to cause self-sinking; human chain-illegal operation driving risky voyage, superimposed on the lack of safety management to exacerbate the risk of loss of control and ultimately lead to sinking. In terms of methodology, the coupling of accident tree and Bayesian network reveals the dynamic mechanism of multi-factors, which provides a new perspective for the study of the causes of complex maritime accidents. The research results can provide a basis for the optimisation of ship safety design, the supervision of high-risk operations and the improvement of navigation system.