Improving Frequency Regulation in Power Systems via Noisynet Deep Reinforcement Learning Approach

The growing utilization of renewable energy resources in contemporary power systems has presented challenges for conventional model-based load frequency control (LFC) methods. These challenges include the escalation of computational burdens and diminished control performance. In order to address the stochastic disturbances, a novel approach called the Noisynet Deep Deterministic Policy Gradient (DDPG) method is proposed to adjust the agent’s parameters. Comparative analysis between the proposed method, the conventional DDPG method, and a finely-tuned PID method demonstrates that the proposed method yields a superior control policy. Consequently, the Noisynet DDPG method shows significant promise as a means to enhance LFC performance.