Research on PCTC accident causal analysis based on association rules and Bayesian network

As the demand for new energy vehicles continues to grow, the number of vehicle exports from our country has steadily increased, and safety issues in large vehicle transport ships (PCTC) have become increasingly prominent, with significant losses caused by accidents. First, through the analysis of 46 PCTC accident investigation reports, 25 causes were identified from four aspects: human factors, machinery, environment, and management. Second, using the Apriori algorithm to uncover the coupling relationships between factors, 50 association rules were generated. Then, based on these association rules, a Bayesian network was constructed, and the BN structure was learned using the K2 algorithm. The BN structure was determined by combining chi-square test results and prior knowledge, and BN parameters were learned using the maximum expectation algorithm. Sensitivity analysis methods were used to extract factors contributing to collisions, sinkings, capsizing, grounding, wind disasters, and fires. The study shows that improper collision avoidance actions, inadequate lookout, and failure to comply with regulations are key factors in collision accidents; improper collision avoidance actions are key factors in sinkings; poor stability and ballast water management issues are key factors in capsizing accidents; lack of supervision and ballast water management issues are key factors in grounding accidents; mechanical failures and ship aging are key factors in wind disaster accidents; electrical failures, flammable material combustion, and insufficient fire extinguishing systems are key factors in fire accidents.