关键词: 交通运输工程；内河危化品运输；运输事故；严重程度；Credal网络模型；IDM

Abstract: To address the problems of data scarcity, cognitive uncertainty and expert subjectivity in the study of inland waterways hazardous chemical transportation accidents, a Credal network model was proposed to analyze and diagnose the severity of inland waterways hazardous chemical transportation accidents, with a view to preventing the risk of accidents from the perspective of safety management. Accident case reports and fault tree analysis were used to identify the risk factors of inland waterways hazardous chemical transportation accidents, and Credal network model based on the parameter learning method of imprecise Dirichlet model (IDM) was constructed to diagnose the severity of inland waterways hazardous chemical transportation accidents. Then, the probability of accidents of different severity levels was predicted through forward reasoning, and the main influencing factors to cause such incidents were identified through diagnostic reasoning. Finally, corresponding safety management measures were proposed according to the results of diagnostic reasoning. The results show that the interval probabilities of accidents occurring in the transportation of hazardous chemicals in inland waterways, categorized as general or lower, moderate, and major or higher, are ［0.711, 0.789］, ［0.172, 0.246］ and ［0.043, 0.069］, respectively. Human factors and management factors are the main factors to cause accidents in the transportation of hazardous chemicals in inland waterways, vessel factors and cargo factors are significant factors exacerbating the severity of these accidents.

Key words: traffic and transportation engineering; inland waterways hazardous chemical transportation; transportation accidents; severity; Credal network model; imprecise Dirichlet model (IDM)