关键词: 贝叶斯网络(BN), 受限空间, 作业情景, 风险分析, 情景构建, 情景推演

Abstract:

In order to accurately identify risks and effectively prevent accidents in confined space operations, a scenario-based accident evolution model grounded in key elements was developed to conduct quantitative risk analysis. This study addressed confined space operations by systematically organizing their key elements, which encompass physical characteristics, spatiotemporal context, environmental conditions, safety management, and emergency decision-making. The Web Ontology Language (OWL) was employed to standardize the description of these elements and represent the scenarios, thereby enabling a structured and standardized characterization of accident evolution. Drawing upon data from nearly 50 domestic confined space accident cases, as well as current standards and expert knowledge, the study performed state-based description and correlation analysis on critical aspects, including gas detection, ventilation, personal protective equipment, and accident consequences throughout the operational process. A BN structure and probability parameters were established accordingly. An empirical analysis of a confined space poisoning accident was conducted to validate the model. The results show that the systematic identification of critical safety factors contributes to the structured management of accident knowledge. Furthermore, the BN model constructed on this basis enables quantitative risk assessment and scenario deduction. The empirical findings demonstrate that risk-ignorant rescue attempts constitute the primary factor exacerbating outcomes in confined space accidents, particularly when multiple non-professional responders enter without personal protective equipment. Therefore, emergency plans should clearly define rescue procedures, responsibilities, and protection requirements.

Key words: Bayesian network(BN), confined space, operations scenario, risk analysis, scenario construction, scenario deduction