Decision-Making Approach for Smart Charging of Electric Vehicles

This paper proposes a cost-effective and user-oriented solution to the problem of smart charging of Electric Vehicles (EVs) in real-time. The proposed approach considers a decentralized framework where the EV user is autonomous to make their own charging decisions in order of minimizing their operating cost. To model the behavior of the EVs under different scenarios, the dynamic programming along with the Markov decision process is adapted. Further, to make the approach respond to a dynamic environment, and learn from historical time series data, the decision tree machine learning models are developed. The feasibility of the proposed smart charging approach is demonstrated by performing offline optimization and testing with the EV data from real-time and numerical simulation sources. The training process of the smart charging approach depicted 96.2% and the testing accuracy is identified to be 98.8%.