A methodology to evaluate human factors contributed to maritime accident by mapping fuzzy FT into ANN based on HFACS

Abstract

A comprehensive accident analysis model is proposed to analyse the human factors that are involved in maritime accidents. The model integrates the superiorities of HFACS, Fuzzy FTA and ANN; HFACS is used to identify and classify human factors associated with 38 accidents, which are re-organized as Basic Events (BEs), Intermediate Events (IEs) and Top Event (TE), developing the architecture of FT; fuzzy aggregation analysis is employed to address expressions of experts, obtaining the failure probabilities of the BEs; and an ANN is established by matching the BEs with variables at the input layer, IEs with the variables in the hidden layers, and the TE with the variable at the output layer. The results show that the proposed ANN-based model has satisfactory performance in overcoming the drawbacks of FTA. The performance of the proposed model can be further improved through the introduction of newly available information/data and the adjustment of the numbers of neurons assigned at the first and second hidden layers according to the architecture of FT. The capability of ANNs-based model to effectively handle uncertainties, dynamics, nonlinear associated with human risk factors is also ascertained through the application for the developed maritime accident scenario.