To enhance the power restoration capability of traction power supply systems under fault conditions, a self-healing reconfiguration and recovery strategy based on a hybrid genetic algorithm and binary particle swarm optimization (GA-PSO) is proposed for traction power supply systems integrated with photovoltaic and energy storage. First, a node-edge equivalent model of the system is established using graph theory. In accordance with the principles of railway fault emergency handling and considering system topology and power flow constraints, a self-healing reconfiguration objective function is formulated with the aims of maximizing the number of restored nodes and minimizing switch operation costs. Subsequently, the breadth-first search algorithm is used to reduce the individual encoding length, and chaotic mapping is introduced for population initialization to enhance the uniformity and diversity of the initial population. And some parameters of the algorithm are improved. Finally, the improved hybrid algorithm is used to solve the self-healing reconfiguration schemes for typical fault conditions in traction power supply systems, and comparisons with other algorithms are made for validation. The case study results indicate that the proposed method achieves excellent convergence speed and optimization capability, significantly improving the fault recovery ability of the traction power supply system.