Reinforcement Learning-Based Plug-in Electric Vehicle Charging With Forecasted Price

摘要: This paper proposes a novel demand response method that aims at reducing the long-term cost of charging battery an individual plug-in electric vehicle (PEV). The problem is cast as daily decision-making for choosing amount energy to be charged in PEV within day. We model Markov decision process (MDP) with unknown transition probabilities. A batch reinforcement-learning (RL) algorithm proposed learning optimum cost-reducing policy from samples and making decisions new situations. In order capture day-to-day differences electricity costs, makes use actual prices current day predicted following Bayesian neural network employed predicting prices. For constructing RL training dataset, we historical linear-programming-based developed creating dataset optimal decisions. Different scenarios are simulated each time frame using set past Simulation results real-world pricing data demonstrate savings 10%–50% owner when method.