MACHINE LEARNING APPLICATIONS IN ELECTRIC VEHICLES

Abstract

Vehicle-to-Grid technology has garnered significant attention due to the increasing number of electric vehicles entering the market. These plug-in electric vehicles can function both as an energy source and as a specific electric load. They can also be used as mobile storage devices to aid in grid load balancing and facilitate the integration of non-conventional energy sources. This thesis focuses on the integration of Electric Vehicles with Vehicle-to-Grid technology under real-time pricing conditions to enhance grid reliability and generate economic benefits for EV owners. The research aims to evaluate the economic viability of using EVs for demand response through bidirectional charging and discharging mechanisms. By employing machine learning models, particularly XGBoost, the study develops predictive models to forecast electricity demand and optimize EV charging and discharging schedules. The methodology includes web scraping for real-time data collection, load forecasting, and RTP analysis to determine the potential revenue generation for different categories of EV users participating in V2G systems.