Enhancing the Sensitivity of Lithium-Ion Battery Parameters via Nonlinear Electrochemical Impedance Spectroscopy

Authors:
Publisher website
Google Scholar
Abstract
A novel multi-modal sensitivity analysis framework is proposed for electrochemical models of lithium-ion battery, integrating time-domain constant-current discharge data with linear and second-order nonlinear impedance spectroscopy data to improve battery model parameter identifiability. By analyzing 26 parameters of a commercial NMC 811 cell (NMC/graphite), we demonstrate that frequency-domain analysis, particularly through the first-ever implementation of second-order nonlinear electrochemical impedance spectroscopy in battery parameter studies, significantly enhances sensitivity quantification capabilities. The framework extends the Doyle-Fuller-Newman pseudo-two-dimensional model deriving frequency-dependent governing equations via Maclaurin/Fourier expansions, which are validated with impedance spectroscopy. Parameter ranges were constrained by experimental calibration, rather than arbitrary intervals, to prevent misleading sensitivity conclusions. The results show that while time-domain methods captured only capacity-related parameters, the frequency-domain approaches improved the identifiability of 14 critical parameters, such as solid-phase diffusion coefficients, double-layer capacitances, and reaction rate constants. Moreover, low-sensitivity parameters such as electrode area and Bruggeman coefficients achieve measurable sensitivity through the nonlinear impedance analysis. The proposed methodology not only quantifies parameter sensitivity but also provides a guideline for developing advanced, multi-modal parameter identification frameworks.
Funding Information
  • National Natural Science Foundation of China (52307238)
  • National Natural Science Foundation of China (52577230)
  • Hebei Province's Yanzhao Golden Peak Talent Recruitment Program for Key Personnel (25JCQNJC01110)
  • Tianjin Natural Science Foundation Youth Science Fund Project (25JCQNJC01110)

This publication has 53 references indexed in Scilit: