The on-board train control system is a core system that ensures the safe and efficient operation of trains, and its failure may cause disruptions to train operations and line congestion. A resilience evaluation of the on-board train control system can provide scientific guidance for optimizing fault-response mechanisms and improving the operational efficiency and stability of regional railway networks. To address this issue, a discrete-time Bayesian network method was adopted. The bidirectional inference capability of the Bayesian network was used to calculate the posterior probabilities of the components in the on-board train control system and identify its weak components. To evaluate the performance recovery capability of the system after a failure, an exponential recovery model was used to quantify the performance curve during the recovery stage. Finally, the resilience triangle area method was used to evaluate the resilience of the identified redundant components. The results show that, when a failure occurs in the on-board train control system, the resilience values of the RTU component using a cold-standby redundancy design and the MT component using a hot-standby redundancy design are 0.9852 and 0.9540, respectively. The resilience values of the redundantly designed components both exceed 0.95, indicating that the redundant structures effectively absorb the impact of failures on component performance and ensure normal system operation.