State of charge estimation of LiFePO₄ battery in AB hybrid battery packs

Breaking the limitations of the conventional single-system battery packs and improving the accuracy of battery state of charge (SOC) estimation are of great significance for the long-term development of new energy vehicles. However, in AB hybrid battery packs incorporating both LiFePO₄ and LiCo_xNi_yMn_1-x-yO₂ cells, the estimation of SOC for LiFePO₄ batteries tends to exhibit significant errors, primarily contributed to the variability of the open circuit voltage (OCV) slope, particularly when the OCV is in the flat region. Therefore, this study proposes to update the SOC estimation algorithm automatically based on the variation of the OCV slope interval and to improve the SOC estimation stability using fusion as well as correction of the initial value. The SOC is estimated in the LiFePO₄ high slope interval using the unscented Kalman filter. During the LiFePO₄ flat-slope interval, the estimation method of the LiFePO₄ cell is mapped using the state of charge of the LiCo_xNi_yMn_1-x-yO₂ cell which can avoid the effect of error accumulation in the LiFePO₄ cell. Furthermore, to accelerate the convergence of SOC estimation, the mapping results are employed to adjust the initial value of the unscented Kalman filter at the end of the discharge period. The results show that the proposed method is able to maintain the SOC estimation accuracy of lithium iron phosphate batteries within 1 %， and the combined algorithm can maintain the maximum error within 4 % under the analogue sensor error.