Demand side management in electric vehicles for smart charging and discharging based on state aggregation and q-learning

Abstract

Use of traditional energy sources results in significant pollution. International organizations are making many efforts to reduce emissions of carbon dioxide (CO2). According to research, EVs can reduce CO2 emissions by 28% by the year 2030. However, the prohibitive price of EVs and the scarcity of outlets for charging continue to be two of the biggest barriers to the widespread use of electric vehicles. In this paper, a detailed demand-side management approach for a network-connected, solar-powered electric vehicle charging station is provided. The proposed approach reduces the requirement for conventional power sources and addresses the current problem of insufficient EVCS by using a solar-powered EVCS in order to make up for the electricity used during peak demand. Models for PV power plants, industrial loads, residential loads, and charging stations for electric vehicles were created using the real-time data. Additionally, a method based on deep learning was devised to control the microgrid’s supply of electricity and to recharge the battery of the electric vehicle during off-peak hours. Two alternative machine learning techniques for figuring out the level of charge in a device that stores electricity were also put to the test. Finally, a 24-h case study using the suggested demand management system structure was conducted. The data show that peak consumption is offset by using a charging station for electric vehicles during peak periods.