An Integrated Framework for Analyzing Risk Influence Factors of Inland Waterway Transport Based on Interpretive Structural Models and Bayesian Networks

Abstract

Historical accident data provides valuable insights into the causes of maritime accidents. To investigate the effect of factors on maritime safety through accident analysis, this study collected 238 accidents that occurred in the mainstream of the Yangtze River from 2016 to 2021. The data features that reflect the frequency of risk influence factors (RIFs) are identified, and principal component analysis (PCA) is used to reduce the feature dimension of the RIFs. Furthermore, an interpretive structure model is constructed to analyze the relevance and hierarchy of the RIFs. The parameters of the network model are learned using the data set of accident cases, and the conditional probability of each node is obtained, based on these, the Bayesian network model of RIFs can be constructed. The sensitivity analysis reveals that all types of accidents are the location of incident, ship type, ship age, hull condition, and channel environment. Four cases are used to verify the effectiveness of the proposed model. This research provides theoretical support for taking measures to prevent accidents and control risks.