Optimized Deep Reinforcement Learning for Smart Charging Scheduling of Plug-in Electric Vehicles

Abstract

In recent decades, fossil fuels have been the major source of energy but concerns about price fluctuations, security, and environmental issues. Plug-in electric vehicles (PEVs) should be connected to the power grid for battery charging. With a wide range of cars, the power system performance in the distribution network. The uncontrolled charging of the battery can cause issues such as overvoltage, overload, unbalanced load, instability, harmonics, and increased loss. Hence, efficient charging methodologies need to be developed for the power grid. The objective of this article is to maximize the total revenue of the charging stations and schedule the charging stations for PEV charging with reduced cost. First, the system model with its objectives such as demand, constraints, and fitness values are computed. Next, efficient charging path planning is proposed using the deep Q network, which is a reinforcement learning model. This can assist the PEV user in selecting the charging station and planning its path. Lastly, the particle sailfish optimization from our previous research work has been used to solve the charging station optimal solution. The simulation results analysis shows the efficiency of the proposed model in reducing the charging cost and time and improves the charging scheduling of PEVs.