Real time prediction algorithm for SOC of lithium ion power battery under high pulse rate

The battery needs to provide a large amount of power in a short time under the condition of a high pulse rate. Real time and accurate State of Charge (SOC) prediction can help the battery management system understand the current status of the battery better, optimize the battery charging and discharging strategy, and improve the efficiency of the battery. In order to prolong battery life and enhance battery safety, a real-time prediction algorithm for SOC of the power battery under a high pulse rate was proposed. The second order RC equivalent circuit is used to establish the model of the battery. The equivalent circuit model of the battery is designed online using the recursive least squares algorithm, and the time-varying parameter model of the battery is established. Its output value is used as the input to the gating recurrent cell neural network, and the neural network is used to output the predicted SOC value. The SOC prediction result is used as the observation vector of the adaptive extended Kalman filter algorithm to obtain the final real-time prediction result of lithium ion power battery SOC. The experimental results show that the parameters identified by the research algorithm for lithium-ion power batteries are as follows: the fluctuation range of ohmic internal resistance is 0.05–0.40 Ω, and the fluctuation range of electrochemical polarization is 0–4.5 F. The terminal voltage values collected by the research algorithm have higher accuracy, with the error being always less than 0.03 V. Moreover, the algorithm can effectively predict the SOC of lithium-ion power batteries in real time, with a maximum average absolute error of about 2%.