Data-Driven Unit Commitment Scheduling Considering Non-Convex Frequency Dynamics

The integration of renewable energy sources into power systems presents significant challenges, including short-term energy supply-demand imbalances and frequency fluctuations. To address these challenges, this paper proposes an improved security-constrained unit commitment model that accurately captures the dynamic frequency response of both inertia-based synchronous generators and converter-interfaced generators. In particular, the pivotal role of energy storage systems (ESS) in providing rapid-response capabilities for frequency regulation is highlighted. a reformulated Markov decision process and a deep reinforcement learning algorithm are developed for optimal unit commitment. A case study on IEEE 39-bus system confirms the effectiveness of the proposed approach in ensuring frequency stability and economic efficiency, simultaneously demonstrating the improvement in frequency dynamics with ESS involvement in frequency regulation.