Closed-Loop Economic Dispatch of Battery Energy Storage Systems Considering Battery Degradation

Battery storage can reduce peak demand and capture the energy arbitrage value, but excessive cycling accelerates battery degradation. This study develops a degradation-aware closed-loop economic dispatch framework for battery energy storage systems (BESSs), where the convolutional neural network–long short-term memory (CNN–LSTM) model is employed as a practical load forecasting module to provide day-ahead operational in-formation. The predicted load is integrated with electricity price, peak-limit constraints, first-order cycling-throughput cost, and terminal state-of-charge (SOC) regulation to generate a baseline charging and discharging schedule. During operation, model predictive control (MPC) updates the remaining horizon using measured load and SOC feedback, enabling closed-loop correction under forecast deviations. The framework is evaluated using Australian load and electricity-price data. On the representative peak-load day, MPC improves terminal SOC from about 43% under fixed scheduling to 49% against a 50% target. Over the annual simulation, the proposed strategy achieves an economic improvement of approximately AUD 20.2 million relative to the no-storage case. With the cycling-throughput penalty included, the modeled equivalent full cycles (EFCs) are reduced from 234.4 to 55.9 compared with the same CNN-LSTM + MPC framework without this penalty. These results demonstrate the value of integrating forecast-guided scheduling, feedback correction, and degradation-aware operation for long-term BESS dispatch.