Research on Personnel Risk Prediction Based on Job Hazard Analysis and Bayesian Network

To accelerate the informatization and systematization of safety management in EMU depot maintenance operations, this paper focuses on first-level maintenance as the research subject and combines Job Hazard Analysis (JHA) with Bayesian Network (BN) for personnel risk prediction. First, the JHA method is used to systematically break down the first-level maintenance process, identifying potential risk factors. Based on these identified risks, a Bayesian Network is constructed to quantify the relationships between the risk factors, predicting the current probability of safely completing first-level maintenance at 99.48%. Finally, through posterior probability and sensitivity analysis, it is determined that the greatest safety impacts are from not wearing protective gear and insufficient EGS discharge time. This method provides a quantitative evaluation approach for the safety management of EMU depot maintenance operations, helping to enhance the risk control capabilities of maintenance personnel.