Energy Storage Management for Wind-Photovoltaic Power Scheduling Compensation Based on Intelligent Combination Estimation of Battery SOC-SOH

Abstract Reliable grid-connected plan implementation can ensure the flexible dispatch capability of new energy. However, it is difficult to achieve due to the unstable output characteristics of wind-photovoltaic power (WPP). The participation of storage in power plan compensation can boost the consumption of new energy, but the function utilization depends largely on reliable energy management. Considering the random alternating conditions of charging and discharging in specific applications, a data-driven self-learning state of charge (SOC) combination prediction model is proposed based on the change characterization of the battery parameters during plan compensation. First, based on historical data statistics of charge/discharge cycle of the battery, the characteristic parameters are extracted such as voltage change, current, and temperature. Then, dynamic SOC and state of health (SOH) prediction combining model is constructed by convolutional neural network-self-attention-long short-term memory (CNN-Self-Attention-LSTM) and the Gaussian process regression (GPR). Second, SOC prediction is modified by introducing SOH dynamic factor to effective management and health assessment of energy storage. Finally, a case simulation is utilized to verify the feasibility of energy storage management in WPP plan compensation. The root mean square error (RMSE), R-squared, mean absolute error (MAE), and mean bias error (MBE) for SOC prediction based on CNN-Self-Attention-LSTM algorithm are 1.513%, 99.675%, 0.59024%, and <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" alttext="negative 0.0098935 percent sign" display="inline" overflow="scroll"> <mml:mo form="prefix">−</mml:mo> <mml:mn>0.0098935</mml:mn> <mml:mtext>%</mml:mtext> </mml:math> , respectively. The accuracy of power scheduling compensation reaches 95.19%. Energy storage can be monitored in real-time to avoid overheating caused by prolonged durations, and prevent major safety accidents. At the same time, the accuracy of the power plan can be ensured.