面向预测不确定性的电池储能系统最优管理：削峰与电池健康协同优化

In modern power grids, integrating renewable energy sources (RESs), deploying battery energy storage systems (BESSs) is increasingly vital for mitigating power fluctuations. However, optimizing BESS operation remains challenging amidst uncertainties in both RES and load forecasting. This paper proposes a novel stochastic model predictive control (SMPC) framework for BESS operation, focusing on peak load shaving and battery health while addressing prediction uncertainties. The proposed framework employs a long short-term memory (LSTM) neural network for forecasting and integrates a constraint-tightening technique into a stochastic optimization (SO) problem with a receding horizon. Based on the load profile of a company in Germany, the proposed framework achieves an additional reduction of 99 kW (5.8%) in peak grid take-out power compared with the traditional model predictive control (MPC) approach, demonstrating its advantage in addressing uncertainties.