Optimization of electric vehicle charging schedule using distributed network computing

Abstract

The significant growth in the number of electric vehicles indicates an increased demand on the power distribution system, specifically on the low-voltage residential network. Without a well-organized schedule for charging electric vehicles, users will typically apply immediate charging upon arrival to home. This may burden the system and may damage power system equipment. To avoid this adverse effect on the system, a process of scheduling electric vehicle charging should be established. This paper proposes a multi-agent based distributed computing process for solving the electric vehicle charge scheduling problem in a secure way that benefits both the customer and the system. This process breaks down the problem into to global and local problem with the former for system objective and the latter for individual vehicle owners' objective. In this work, the local problems are modeled as sub gradient problems that can be solved simultaneously by corresponding agents. The optimality of the sub gradient solutions with respect to global objective are made sure through information sharing between the agents during each iteration. The detailed modeling and implementation of the proposed method along with numerical analysis to demonstrate the effectiveness are presented in the paper.