LAUSR

Abstract To improve the estimation accuracy of a battery’s inner state for a battery management system, an improved online model-based parameter identification algorithm is proposed. To reduce the computation cost, the existing methods regard the open circuit voltage over a certain time as a constant value. However, the battery state-of-charge (SoC) estimation error with the traditional method will deteriorate with larger sampling intervals. Compared with the existing parameter identification method, a new online estimation method is proposed, and both recursive least squares (RLS) and least mean square (LMS) algorithms are employed and compared systematically. The LMS algorithm, which requires less computational capability and storage space but performs worse than the RLS algorithm, is also invalid for the wide sampling interval in the traditional method. The improved method using LMS can maintain the maximum SoC estimation error at less than 10%. The simulation results show that the proposed approach can accurately identify the model parameters within 5% SoC estimation error. Finally, a hardware-in-the-loop validation experiment is carried out to prove the accuracy and superiority of the improved method.