Learning-Aided Adaptive Lyapunov Optimization for Wind Farm-Equipped Storage Control

With the growing integration of wind power into the energy supply, the grid's stability faces heightened challenges due to the unpredictable nature of wind energy. For the wind farms, additional electricity penalty costs will be charged because of the real-time discrepancies between the committed and actual supplied wind power. Equipping wind farms with energy storage systems can help alleviate these power discrepancies. However, implementing an optimal online storage control policy is practically challenging due to time-varying penalty costs and energy mismatches. To address this issue, we propose a storage control algorithm based on the Lyapunov optimization frame-work. To further improve the performance of the time-invariant Lyapunov function, we employ a deep neural network (DNN)-driven parameter-learning approach and design a novel time-adaptive Lyapunov optimization framework. Simulation results demonstrate the effectiveness of our proposed algorithm.