Advanced Load Scheduling in HEMS with Deep Reinforcement Learning

The Smart Grid marks a significant advancement in modern energy infrastructure, offering improved efficiency, reliability, and sustainability. A key component of the Smart Grid's success is the integration of Home Energy Management Systems (HEMS), which are essential for optimizing energy con-sumption, reducing peak loads, and facilitating demand response. HEMS enable intelligent and adaptive control of residential energy usage, contributing to the overall stability, resilience, and sustainability of the Smart Grid. This paper presents a comprehensive study on optimizing HEMS with a focus on load scheduling and cost minimization. Using the REFIT dataset, existing literature has explored various mathematical, meta-heuristic, and machine-learning algorithms. However, this study introduces a novel approach by leveraging Deep Reinforcement Learning (DRL) with Double Deep Q-Learning (DDQL) and compares its performance with Deep Q-learning (DQN). The experimental results show that the proposed DRL algorithm consistently outperforms the DDQL method in achieving optimal home energy management, indicating promising advancements for efficient HEMS applications.