Optimal Scheduling of Electric Vehicle Charging Using Intelligent Control Strategies

The growing adoption of electric vehicles (EVs) causes peak load spikes, voltage instability, and more as grid operating costs, thereby requiring intelligent charging coordination. This chapter offers a MATLAB/Simulink framework for EV optimal charging operation based upon reinforcement learning (RL) algorithms, namely Deep Q-Network (DQN) and Proximal Policy Optimization (PPO). A realistic Simulink-modelled microgrid consisting of renewable generation, load, and charging stations trains the RL agent using the MATLAB Reinforcement Learning Toolbox to make charging policies that minimize energy cost, flatten load with a low peak-to-average ratio (PAR), and utilize renewables as much as possible. Design of reward function, training agents, and convergence analyses are covered and compared with linear programming and heuristic methods. Results of the simulation indicate greater benefits of savings, load flattening, and proper charging, hence the support for the UN Sustainable Development Goal 7 towards affordable and clean energy.