Abstract
This research aims to develop a reinforcement learning (RL)-based control strategy to manage imbalance risks through the aggregation of residential flexible assets. The proposed strategy is designed to efficiently operate within sequential electricity markets, addressing uncertainties in household energy consumption while enhancing grid stability, reducing operational costs, and offering financial savings for consumers. The research hypothesizes that RL approach will provide significant improvements in cost savings and decision-making efficiency compared to traditional optimization methods. Furthermore, it aims to increase the transparency of RL-based methods through explainable strategies to foster user trust and promote widespread adoption. The study consists of three objectives: (1) defining a comprehensive market framework that captures the dynamics of electricity markets and quantifies the economic impact of imbalance prices on Balance Responsible Parties (BRPs) and consumers; (2) developing RL-based strategies to manage imbalance risks and optimize energy arbitrage; and (3) validating the proposed control strategy through real-world pilot studies, assessing its effectiveness and interpretability to various household configurations. By combining state-of-the-art RL techniques with real-world deployment, this research will offer a scalable, efficient, and transparent solution for optimizing residential flexibility and enhancing grid operations.
Researcher(s)
Research team(s)
Project type(s)