Research on Improving the Operational Efficiency of Battery–CAES Systems Using a Dual-Layer Optimization Model Based on CNN-LSTM-AM Forecasting

This study addresses the low operational efficiency and high energy storage cost of wind–solar hybrid energy storage systems due to the strong volatility and intermittency of wind and photovoltaic power. Instead, the authors propose a dual-layer optimization model based on convolutional neural network–long short-term memory–attention mechanism (CNN-LSTM-AM) forecasting. First, a CNN-LSTM-AM forecasting model is constructed based on convolutional neural networks and long short-term memory networks. Then, the model is applied to wind and solar power forecasting to dynamically optimize the output power ratio of renewable sources and batteries based on predicted power, thereby reducing the start–stop frequency of compressed air energy storage (CAES) and improving operational efficiency. For lower-layer optimization, a weight evaluation model based on AHP is constructed and subsequently used to optimize the capacity configuration of the hybrid energy storage system to achieve overall system optimality. Case studies indicate that after upper-layer optimization, the number of CAES start–stop cycles decreases from 25 to 17, and further declines to 14 after optimization of the lower-layer capacity configuration, while the energy storage cost is reduced by 5.43% and the curtailment rate decreases by 0.15%. This validates the effectiveness of the proposed model in improving the economic performance and stability of renewable hybrid energy storage systems.