Al-based Peak Load Reduction Approach for Residential Buildings using Reinforcement Learning

In recent years, energy management has gained significant interest from industries and the research community. The goal of this paper is to explore the Artificial Intelligence (AI) for a residential building to reduce peak load using Reinforcement Learning (RL). Denning an Al-based system improves the likelihood to save energy and reduce burden on Smart Grid (SG) during peak hour along with computational aspects of RL. Here, we proposed a Q-learning-based Peak Load Reduction (QL-PLR) approach to provide an optimum solution for Residential Energy Management (REM). Further, to handle uncertainties, a scenario-based mechanism is used with the real data along with probability density functions. Deterministic numerical calculations are done to justify the efficacy of the proposed QL-PLR approach. The experimental results indicate that the application of the proposed QL-PLR approach leads to lower energy consumption associated with consumer dissatisfaction compare to existing approaches with respect to different evaluation metrics.