Abstract: Maintenance strategy optimization with Bayesian network analysis based on fault tree and condition-based maintenance model is mainly studied. After obtaining the fault data of metro signaling system on-board equipment, the Bayesian network analysis is adopted to study the identification method of signaling system on-board equipment importance degree. On the basis of periodic preventive maintenance, the condition-based maintenance model is used to carry out targeted optimization, achieving the goal of risk and cost dual optimization, strengthening weak parts of signaling system on-board equipment key component.