Optimal Energy Management Strategy for Smart Home with Electric Vehicle
Yu Yi, Gregor Verbič, Archie C. Chapman
Abstract
The increasing adoption of electric vehicles (EVs) poses a challenge to the operation of the electricity grid, but also offers opportunities. In this paper, we propose an energy management framework for a smart home that can operate both in a vehicle-to-grid (V2G) and vehicle-to-home (V2H) modes, thus simultaneously minimizing energy expenditure for the homeowner and mitigating network load at peak hours. The home energy management problem is formulated as a Markov decision process (MDP) with the objective of household electricity cost minimization considering the randomness of the EV mobility, photovoltaic (PV) generation, and household demand. We use reinforcement learning based on the deep deterministic policy gradient (DDPG) algorithm to determine the optimal charging policy for the EV and the battery storage system and compare it to a policy obtained using deterministic mixed-integer linear programming (MILP). The simulation results demonstrate the effectiveness of using the EV for home energy management.