State-of-charge estimation combination algorithm for lithium-ion batteries with Frobenius-norm-based QR decomposition modified adaptive cubature Kalman filter and H-infinity filter based on electro-thermal model

A novel algorithm containing an adaptive cubature Kalman filter (ACKF) modified by Frobenius-norm-based (fro-norm-based) QR decomposition (QR) and H-infinity(H∞) filter based on electro-thermal model is proposed to estimate the state of charge (SOC) of lithium-ion batteries (LIBS). First, an electro-thermal model with a second-order RC equivalent circuit model (ECM) and a lumped thermal model is employed to identify the internal parameters of LIBS at different temperatures. Then, to solve the non-positive definiteness of the error covariance matrix, an adaptive cubature Kalman filter is modified by fro-norm-based QR decomposition (ACKF-QR). Finally, to cope with uncertain noises especially non-Gaussian noises, the H∞ filter is combined with ACKF-QR to estimate the battery SOC (ACKF-QR-H∞). The ACKF-QR-H∞ algorithm is validated under different working conditions at different temperatures with incorrect initial values. The SOC estimation MAXE (Maximum absolute error) of the ACKF-QR-H∞ algorithm is less than 1% and its SOC estimation MAE (Mean absolute error) and RMSE (Root mean square error) are less than 0.32%. As compared with the same algorithm without considering temperature variations, the SOC estimation error of ACKF-QR-H∞ algorithm can almost reduce by half in most cases. When various noises are added manually, the ACKF-QR-H∞ algorithm can remain robust.