Data on Machine Learning regenerated Lithium-ion battery impedance

Abstract

This paper describes and provides the data on the regenerated-impedance spectra that is computed from experimental results of electrochemical impedance spectroscopy measurements taken from a commercial Li-ion battery. The empirical impedance data of secondary coin type Li-ion batteries were collected in different states of charge ranging from empty to full state of charge configurations. This approach utilizes only a small seed (ex grano) experimental data set to first build an ensemble of weighted disparate models selected based on performance and non-correlative criteria (“co-modelling”) then second to generate what would be the remaining experimental data synthetically. The “Cooperative Model Framework” demonstrates the efficacy of this approach by assessing the synthetically generated data.

Keywords

• Machine Learning (ML) on Li-ion batteries
• Co-modeling approach
• Electrochemical Impedance Spectroscopy (EIS) for Li-ion batteries
• Regeration of impedance for Li-ion batteries