A novel square root adaptive unscented Kalman filter combined with variable forgetting factor recursive least square method for accurate state-of-charge estimation of lithium-ion batteries.

Lithium-ion battery state-of-charge (SOC) serves as an important battery state parameter monitored by the battery management system (BMS), real-time and accurate estimation of the SOC is vital for safe, reasonable, and efficient use of the battery as well as the development of BMS technology. Taking the ternary lithium battery as the research object, based on the second-order RC equivalent circuit model, a variable forgetting factor least square method (VFFRLS) is used for parameter identification and a combination of the square root of covariance and noise statistics estimation techniques to estimate the SOC, to solve the problem of dispersion of the unscented Kalman filter and the error covariance tends to infinity with iterative calculation, thus ensuring the accuracy of SOC estimation. The feasibility and robustness of the algorithm and the battery state estimation strategy are verified under HPPC and BBDST conditions with maximum errors of 1.41% and 1.53%, respectively. The experimental results show that the combined algorithm of VFFRLS and SRAUKF has good robustness and stability, and has high accuracy in the SOC estimation of Li-ion batteries, which provides a reference for the research of lithium-ion batteries