A Hybrid Optimization Framework for Equivalent Circuit Model Parameter Identification from Electrochemical Impedance Spectroscopy of Low-Impedance Lithium-Ion Battery Cells for Electric Vehicles

Electrochemical impedance spectroscopy is widely used for lithium-ion battery development, diagnostics, performance evaluation, and state-of-health and state-of-charge estimation. However, the characterization of equivalent circuit models from experimental data remains largely dependent on a trial-and-error process. The fitting of model parameters from experimental data are a nonlinear, nonconvex optimization problem with multiple local minima. The conventional approach often requires carefully selected search bounds and iterative manual adjustment of initial guesses. The problem is more challenging for low-impedance electric vehicle battery cells, whose impedance spectra show complex diffusion phenomena, inductive effects, and higher measurement noise. To address this issue, this work presents a robust and automated hybrid global-to-local optimization framework for equivalent circuit model parameter identification using a genetic algorithm followed by Nelder–Mead refinement. The electrochemical impedance spectroscopy data reported in a previous study are adopted as a benchmark to evaluate this framework. The fitting results across different states of charge and temperatures demonstrate high convergence reliability, high fitting accuracy, robustness to initial conditions and search bounds, and computational efficiency. This proposed workflow provides a practical and reliable framework for equivalent circuit model parameter identification from electrochemical impedance spectroscopy of milliohm-level lithium-ion battery cells used for electric vehicles. Across six state-of-charge–temperature conditions, the genetic algorithm reached stable near-optimal solutions within approximately 150 iterations and final residual values ranging from 2.14×10−3 to 2.89×10−3. The following Nelder–Mead refinement achieved at most 0.012% additional residual value. In general, this workflow provides a practical and reliable approach for equivalent circuit model parameter identification from electrochemical impedance spectroscopy data and can be extended to a wider range of battery systems.