Wayside storage regenerative braking energy recovery system can effectively absorb and reuse the residual regenerative braking energy of trains

and has been widely used in urban rail transit. This paper proposes an energy management strategy based on improved deep deterministic policy gradient algorithm for the characteristics of the traction power supply system in the actual scenario of multi-substations and multi-energy storage systems (ESSs). On the one hand

this strategy solves the continuous control problem proposed by the efficient utilization of regenerative braking energy for the ESS in the actual operation condition; On the other hand

fuzzy logic control is used to guide the agent learning to solve the reward shaping difficulties and sparse rewards when applying deep reinforcement learning. In addition

the proposed strategy is verified by simulation based on actual line conditions

and the results show that the proposed strategy can significantly optimize the control effect of the ESS to achieve better energy saving and voltage regulation.