Dynamic risk assessment for train brake system considering time-dependent components and human factors

The train brake system (TBS) is a critical segment for the safety of train system whose failure can lead to serious consequences. Due to the dynamic and multi-factors coupling characteristics of TBS failure, it is important to construct a comprehensive risk assessment method to capturing the time-dependent and human factors. This paper proposes a model that incorporates degraded components and human factors into dynamic Bayesian networks to formulate dynamic risk assessment for the TBS. In the proposed method, FT is employed to determine the logical structure of the failure process. DBN is used to capture the dynamic features of TBS failure in which the human factors are modelled by the CREAM method and the degraded components are formulated by MC. Meanwhile, different maintenance strategies are considered in the proposed hybrid method. In particular, we evaluate the degradation of four common failures, that is, the insufficient braking, brake test failure, brake release failure, and wheel lock. The risk-influencing factors of the brake system and their relevance are identified. When considering the cause of brake system failure, we analyse the human factors in the system through the reliability analysis method. The results show that the proposed method is able to capture the spatial variability of parameters and simulates the evolution of brake faults in time and space domains. Some practical suggestions are provided to the operators of the system based on the sensitivity analysis.