Uncertainty analysis of accident causality model using Credal Network with IDM method: A case study of hazardous material road transportation accidents

Bayesian network (BN) is an effective tool for causal inferences of accidents. However, it is often criticized for the difficulty in obtaining accurate/sufficient data needed to get precise probability numbers, and expert knowledge is necessary on this occasion. Such numbers provided by experts lead to great cognitive uncertainty in the model. Credal Network (CN), regarded as an extension of BN, uses imprecise probabilities to well express the cognitive uncertainty, but how to give the available interval probability is a key issue in its application. In this paper, a CN model based on Imprecise Dirichlet Model (IDM) is proposed to solve this problem, and a case study of hazardous material road transportation accidents (HMRTAs) is conducted. Firstly, based on the accident causation model, the CN topology is established by analyzing 30 investigation reports of HMRTAs in China. Secondly, the conditional probability table is calculated and then the CN model of HMRTAs is obtained by transforming the point probability into interval probability using IDM. Finally, causes and consequences of HMTRAs are analyzed. The results show that the method in this study can well represent the uncertainty propagation in the accident causal model and provide a new way for accident risk analysis. (c) 2021 Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

Automated summary

This paper proposes a Credal Network model based on Imprecise Dirichlet Model for addressing cognitive uncertainty in accident causal inference, with a case study on hazardous material road transportation accidents. The results demonstrate that the method can effectively represent uncertainty propagation in the accident causal model and offer a new approach for accident risk analysis.