State of Charge and State of Temperature Estimation Method for Secondary Batteries Using Neural Networks and Electrochemical Impedance Spectroscopy

Abstract

This study proposes a method to estimate the states-of-charge (SoCs) and states-of-temperature (SoTs) of secondary batteries using neural networks and electrochemical impedance spectroscopy. The impedances in the frequency range of 100mHz-10kHz of a general lithium-ion battery were measured for various SoCs and SoTs and used for training a neural network with two hidden layers. The performance was evaluated using the measured impedances that were not used for the training. The mean square errors obtained were 2.094% and 0.511°C for SoC and SoT respectively.