Data-Driven Policy Mapping for Safe RL-based Energy Management Systems
This addresses energy management challenges for building operators by offering a scalable and safe solution, though it is incremental as it combines existing techniques like clustering and forecasting with RL.
The paper tackles the problem of scalable and safe energy management in buildings by proposing a reinforcement learning-based system that clusters load profiles, forecasts future states, and uses action masking, resulting in up to 15% cost reduction and stable performance for new buildings without retraining.
Increasing global energy demand and renewable integration complexity have placed buildings at the center of sustainable energy management. We present a three-step reinforcement learning(RL)-based Building Energy Management System (BEMS) that combines clustering, forecasting, and constrained policy learning to address scalability, adaptability, and safety challenges. First, we cluster non-shiftable load profiles to identify common consumption patterns, enabling policy generalization and transfer without retraining for each new building. Next, we integrate an LSTM based forecasting module to anticipate future states, improving the RL agents' responsiveness to dynamic conditions. Lastly, domain-informed action masking ensures safe exploration and operation, preventing harmful decisions. Evaluated on real-world data, our approach reduces operating costs by up to 15% for certain building types, maintains stable environmental performance, and quickly classifies and optimizes new buildings with limited data. It also adapts to stochastic tariff changes without retraining. Overall, this framework delivers scalable, robust, and cost-effective building energy management.