Advanced Home Energy Management Using Proximal Policy Optimization with a Comprehensive Appliance Set

Home Energy Management Systems (HEMS) are essential for optimizing household energy consumption and reducing costs, particularly in smart grids, where renewable energy sources and demand side management play a critical role. We propose an advanced HEMS framework that utilizes Proximal Policy Optimization (PPO), a reinforcement learning (RL) algorithm to address the challenges of managing energy consumption in realistic and dynamic environments. Our approach provides a comprehensive smart home environment by incorporating a wide range of household appliances, each with distinct patterns of energy consumption and operational constraints. This enhances the realism and practical relevance of the system, allowing cost savings while respecting user preferences and the limitations of the appliance. Simulations reveal that the proposed HEMS framework significantly improves energy savings, reduces costs, and enhances user satisfaction compared to other baseline methods, achieving 38% lower costs and 3% higher satisfaction of the energy level of electric vehicles (EV) than the Soft Actor-Critic based HEMS. These results highlight the effectiveness of RL and realistic environment modeling in the development of adaptive and efficient HEMS solutions, paving the way for more sustainable energy management practices.