Scholarlink

<p>Accurate photovoltaic (PV) forecasting and multi-scale dynamic dispatch of energy storage systems are crucial for ensuring stable microgrid operation and optimizing energy management. This study focuses on a microgrid integrating distributed photovoltaics with a hybrid energy storage system (HESS) composed of vanadium redox flow batteries, lithium-ion batteries, and supercapacitors. A novel optimization strategy for HESS dispatch is proposed, which combines a hybrid neural network model for short-term PV forecasting with a variable-domain fuzzy inference system incorporating an adaptive feedback mechanism. First, a short-term PV forecasting model (SSA-CNN-LSTM) is developed by optimizing the convolutional neural network (CNN) and long short-term memory (LSTM) network parameters using the sparrow search algorithm (SSA). By integrating multiple influencing factors, the model achieves high-precision short-term PV power prediction, providing a reliable power reference for subsequent fuzzy inference. Subsequently, the frequency decomposition (partial filtering) of the forecasted PV power fluctuations is combined with an adaptive feedback mechanism using variable-domain fuzzy inference. This approach enables rational power distribution among the three types of energy storage systems—high-frequency, medium-frequency, and low-frequency—while maintaining overall energy balance of the HESS. Finally, simulations conducted in MATLAB/Simulink verify the effectiveness of the proposed HESS dispatch optimization strategy for the microgrid.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Combining SSA-CNN-LSTM photovoltaic prediction model with variable theory domain fuzzy theory for microgrid adaptive scheduling optimization of hybrid energy storage system with vanadium flow batteries

  • 期刊论文

  • 发表日期：

    2026年3月2日

查看全文
  • Jianbin Li,
  • Zhengxiang Song

摘要

Accurate photovoltaic (PV) forecasting and multi-scale dynamic dispatch of energy storage systems are crucial for ensuring stable microgrid operation and optimizing energy management. This study focuses on a microgrid integrating distributed photovoltaics with a hybrid energy storage system (HESS) composed of vanadium redox flow batteries, lithium-ion batteries, and supercapacitors. A novel optimization strategy for HESS dispatch is proposed, which combines a hybrid neural network model for short-term PV forecasting with a variable-domain fuzzy inference system incorporating an adaptive feedback mechanism. First, a short-term PV forecasting model (SSA-CNN-LSTM) is developed by optimizing the convolutional neural network (CNN) and long short-term memory (LSTM) network parameters using the sparrow search algorithm (SSA). By integrating multiple influencing factors, the model achieves high-precision short-term PV power prediction, providing a reliable power reference for subsequent fuzzy inference. Subsequently, the frequency decomposition (partial filtering) of the forecasted PV power fluctuations is combined with an adaptive feedback mechanism using variable-domain fuzzy inference. This approach enables rational power distribution among the three types of energy storage systems—high-frequency, medium-frequency, and low-frequency—while maintaining overall energy balance of the HESS. Finally, simulations conducted in MATLAB/Simulink verify the effectiveness of the proposed HESS dispatch optimization strategy for the microgrid.