关键词:  , 关键词： 钒电池, 荷电状态（SOC）估算, 卡尔曼滤波算法, BP神经网络, 储能

Abstract:  ABSTRACT：   Traditional Kalman filter method on the SOC estimation of vanadium redox flow battery regards model parameters as constant. In view of that fact which leads to error increasing, this paper uses BP （back propagation） neural network to online update the real-time parameter values of the Kalman filter process to achieve higher accuracy. By using the common Thevenin equivalent circuit model, the internal resistance R0, polarization resistance Rp and capacitance Cp are updated through neural networks to complete the Kalman filter optimization so that the system model is updated in every step of Kalman filter estimation to make up for the defects of the traditional algorithm. Meanwhile, the battery test experiment is designed. Compared with the result of dual Kalman filter optimization, the results indicate that the method of neural network optimization can better reflect the dynamic characteristics of the system, and the estimated result has higher accuracy and better convergence. It is proved that this method is suitable for real-time SOC estimation of vanadium battery system and has practical significance and application value.

Key words:  KEYWORDS：  vanadium redox flow battery, SOC estimation, Kalman filter algorithm, BP neural network, energy storage