Incorporating public feedback in service restoration for electric distribution networks

Power outages in urban area carry heavy social and economic costs. Although social cost, especially public sentiment, is concerned by engineers and managers, it has been only qualitatively investigated without a rigorous model in the state-of-the-art research and practice of service restoration (SR) for a long time. To fill this gap, this paper investigates a hybrid model which takes public sentiment into consideration by quantifying public sentiment triggered by power outage. Furthermore, conventional SR method focused on the optimization model with ideal conditions, which leaves a large room for improvement in complex environment. To improve the robustness of the model, the authors propose a reinforcement learning framework to analyze emergency management process without prior rules. At each time step, the optimal decision can be made automatically by a learned model. The numerical simulations with modified IEEE 33-bus and IEEE 123-bus systems demonstrate the effectiveness of the proposed method.