Online Distributed Gradient Projection based Charging Strategy for EVs in Residential Areas

In this paper, we investigate a demand response management scenario in residential changing station where electric vehicle (EV) users, who have privacy concerns and may charge EV unorderly, compete to minimize their charging cost. Based on game theory, we adopt an hourly billing model using dynamic pirce to mitigate the harm caused by unordered charging of EVs. Considering the uncertainty of users' behavior, we propose an online distributed Nash equilibrium (NE) parallel calculation method based on gradient projection descent, which can address the concern of users' privacy, timely respond to load changes and support a large number of EVs. The case study shows the advantages of the method in valley-filling of the load profile and the online algorithm can maintain the user cost near the cost of predicted cases.