The multi-innovation adaptive extended Kalman filter algorithm for battery SOC estimation

In order to solve the problem of low accuracy of online estimation of lithium battery under unknown noise, an adaptive extended Kalman filter algorithm combined with the theory of multi-innovation was proposed. On this basis, in order to verify the validity of the algorithm, firstly established for MI - AEKF second order battery model to estimate the SOC(state of charge) by using the parameter identification method of offline recognition model of relevant parameters and model precision. Secondly, experiments are designed to compare the estimation accuracy of multi-innovation extended Kalman filter method with traditional EKF and AEKF methods, so as to realize the accuracy verification of MI-AEKF method.The experimental results show that the error of the MI-AEKF method in estimating SOC under unknown noise is less than 1.8%, while EKF and AEKF fluctuate between 2% and 3%. Compared with EKF and AEKF, the MI-AEKF method has the advantages of fast convergence speed and high estimation accuracy.