A Novel Data-Driven Optimization for Cost-Effective Home Energy Management with PV-EV Integration

As electric vehicle (EV) adoption accelerates; it increases power grid demand but also offers significant benefits. EVs can act as energy storage units, managing grid load and facilitating renewable energy use. Integrating renewables like photovoltaic (PV) systems with EVs through a Home Energy Management System (HEMS) eases grid stress and reduces carbon emissions. In this study, we present a novel Data-Driven-Model (DDM) based on the pioneer Proximal Policy Optimization (PPO) algorithm, a recent advancement in machine learning. The aim is to optimize policy formulation in sequential decision-making tasks in HEMS, specifically to minimize household monthly electricity costs. This approach leverages a grid-assisted bidirectional PV-EV system in Sydney, Australia households, operating in both vehicle-to-grid and vehicle-to-home modes. The performance of this method is comprehensively assessed and compared with the base case, demonstrating a significant reduction in monthly electricity costs.