Machine Learning-Based Day-Ahead Energy Forecasting and PSO-Based Coordinated Charging of Electric Vehicles in a Residential Complex

Electric vehicles (EVs) offer significant benefits in reducing emissions and oil consumption, but their widespread adoption depends on adequate charging infrastructure. Residential complexes, with numerous parked cars, are ideal for establishing such infrastructure. However, optimally distributing limited charging power, constrained by transformer capacity and overall power usage, poses a significant challenge. This paper presents a novel approach for day-ahead energy forecasting and coordinated EV charging in residential complexes using machine learning and optimization techniques. Our method predicts next-day energy requirements for each EV by leveraging historical data, EV physics models, and machine learning algorithms. Based on these forecasts and available power, optimization techniques are applied to efficiently distribute power among the EVs. The effectiveness of the proposed approach is validated through simulations and real-time data collection, demonstrating its potential to optimize EV charging in residential settings.