Load Frequency Control Enhancement Using Reinforcement Learning Technique

Abstract

Microgrids (MGs) face challenges due to load disturbances, the uncertain nature of renewable output power, energy storage system dynamics, and low system inertia. These factors can lead to large frequency deviations, weakening the MG and potentially resulting in a complete blackout. Addressing this, this paper introduces a load frequency control (LFC) method against stochastic power flow from renewable energy sources, leveraging deep reinforcement learning (DRL). A real-time MG test system is employed for simulation purposes. This system is modeled using MATLAB/Simulink, and its performance under various scenarios is analyzed to evaluate the efficacy of the proposed method, contrasting it with existing techniques from the literature. Results indicate that our proposed controller offers a more rapid response and is well-suited for dynamic systems.