Human Reliability Analysis for Inland Waterway Navigation

1 Inland waterway transport (IWT) is an important part of waterway transport. Prevention and 2 reduction of inland waterway traffic accidents is always the most central issue of developing IWT. 3 And human errors contribute the most for ship accidents. As a dynamic process, the ship accident 4 results from a series of factors which are related in a complicated manner. This paper presents a 5 general method for developing a Bayesian Network to model the human reliability analysis of 6 inland waterway ship accidents. The qualitative part of the BN is structured based on three expert 7 group. The quantitative part is based on correlation analysis of human related factors from 123 8 accident investigation reports. The Bayesian inference using GeNIe software shows that improper 9 shift and poor lookout have a significant effect on the probability of ship accidents. The main 10 causation factors could be found out and analyzed through BN probabilistic reasoning, thus the 11 proposed BN model provides effective suggestions for decision making about inland waterway 12 safety related issues. 13 14 keywords: Inland Waterway Transport, Navigation Safety, Bayesian Network, Expert Elicitation, 15 EM Algorithm 16 17