An Improved Multi-Time Scale Lithium-Ion Battery Model Parameter Identification Algorithm Based on Discrete Wavelet Transform Method

Abstract

Efficient battery management system (BMS) monitoring and accurate battery state estimation are inseparable from precise battery models and model parameters. Because of the multi-time scale dynamic characteristics of the battery system, there are still challenges in the modeling and parameter identification accuracy of the battery equivalent circuit model (ECM) in this case. This article proposes a multi-time scale parameter identification algorithm based on multiresolution analysis (MRA) of discrete wavelet transform (DWT), which is used for closed-loop estimation of battery ECM parameters corresponding to different electrochemical dynamic effects. The ECM of the battery at multiple time-scales is determined by the distribution of relaxation times (DRTs) method, and MRA decomposition is performed on the battery signal to determine the separated and decoupled model parameters. The open-circuit voltage (OCV) is used as a slow time-scale model parameter and does not require offline state-of-charge (SOC)-OCV calibration. Under the urban dynamometer driving scheme (UDDS) experiment, the estimation results of ECM parameters, terminal voltage, and SOC using the proposed algorithm were compared with those obtained using different implementation methods. The root mean square error (RMSE) results show that the algorithm can accurately estimate the terminal voltage, OCV, and SOC of the battery, with estimation errors of 0.966, 2.58mV, and 0.1263%, respectively.