Multi-Area Coordinated Control of High-Proportion New Energy Power Grids Based on an Improved Deep Deterministic Policy Gradient Algorithm

Driven by the “dual carbon” goals, the installed capacity of new energy power generation has significantly increased, with its share in total installed capacity and power generation markedly rising, gradually transitioning from supplementing electricity supply to becoming the primary power source. However, as new energy is increasingly integrated into the grid on a large scale, the growing randomness and intermittency of the power system create substantial challenges for maintaining its safe and stable operation. This paper introduces a coordinated control algorithm for multi-area interconnected grids, specifically the KL-divergence penalized Twin Delayed Deep Deterministic Policy Gradient algorithm. The proposed algorithm utilizes dual Q-networks along with delayed policy updates and target policy smoothing regularization to mitigate Q-value overestimation. Additionally, it incorporates a KL-divergence penalty to regulate the extent of policy updates, which in turn enhances the stability of the training phase. Simulations based on the IEEE two-area load frequency control model show that the proposed algorithm greatly improves system control performance, enabling coordinated control over various regions.