A NOISE COMPENSATED METHOD FOR MODEL PARAMETERIZATION OF LITHIUM‐ION BATTERY

A well‐parameterized battery model is prerequisite of the model‐based estimation and control methods. This paper focuses on the unbiased model parameter identification when noises corrupt the measurements. The parameter identification problem within the noise corruption scenario is reformulated as a nonlinear least squares (NLS) problem. A novel offline two‐step method combining least squares (LS) regression and variable projection algorithm (VPA) is then proposed to co-estimate the noise variances and unbiased model parameters. The proposed LSVPA is further extended to the online recursive version by using the Gauss‐Newton (GN) method. Simulation and experimental results show that the proposed method can well compensate for the noise effect and improve the accuracy of model parameterization.