关键词: SOC, 递推最小二乘法, IMM-UKF, 马尔科夫链, 自适应性

Abstract:

In the prediction method of the power lithium battery state of charge （SOC）, there are problems such as the cumulative error of the ampere-time integration method and the divergence of the estimation result of the extended Kalman filter algorithm. This paper proposes a soc estimation strategy based on the interactive multi-model unscented Kalman filter （IMM-UKF） algorithm. Firstly, the second-order RC battery equivalent model is established，The recursive least squares method with forgetting factor is used to identify the battery equivalent model parameters online, and consider the battery's actual capacity change and sensor noise caused by the discharge of the battery under different magnification conditions. Three different parameters of the battery model of current, medium current and small current, then study the Markov chain between the three models, determine the transition probability and model probability between each model based on the prior information, and finally build the matlab simulation model，The experimental results show that the average error of IMM-UKF is less than 1%, the adaptability of the algorithm is enhanced, and the prediction accuracy is improved, which has better prediction effect than the current mainstream prediction methods.

Key words: SOC, push the least squares method, IMM-UKF, Markov chain, adaptive