Estimation Method of Solid Phase Diffusion Time-constant of Lithium-ion Battery Based on Time-domain Data of Two-electrode Battery and Neural-network

The solid phase diffusion (SPD) time constant (SPD-TC) is an important parameter of the fractional-order model (FOM) of the lithium-ion battery. The estimation of SPD-TC is one of significant problem for the battery aging analysis. Both a three-electrode pouch cell with a copper micro-reference electrode and the electrochemical impedance spectroscopy (EIS) are employed to identify the SPD-TC recent years but the method is hardly used in the field of engineering applications. This paper proposes a method to estimate the SPD-TC based on time-domain data of two-electrode battery and a back propagation neural-network (BPNN). First, the FOM of the battery is adopted to generate samples to train a BPNN. Then, the terminal voltage of a ternary lithium-ion battery, measured by a step current discharge experiment, is used as the input of the trained BPNN, and SPD-TCs of positive and negative electrodes are estimated. Finally, the error is analyzed according to the EIS of the positive and negative electrodes.