Robustness assessment and enhancement of deep reinforcement learning-enabled load restoration for distribution systems

Haipeng Xie, Lingfeng Tang, Hao Zhu, Xiaofeng Cheng, Zhaohong Bie

Abstract: Highlights•Evaluate the robustness of deep reinforcement learning-enabled load restoration for distribution networks.•Establish an adversarial example generation model for the load restoration agent.•Mitigate deleterious impacts of state perturbations on critical load supply using adversarial training.