Artificial Intelligence-based Battery State-of-Health (SoH) Prediction through battery data characteristics analysis

Abstract

Batteries are used in various places, including portable devices and energy storage devices. However, due to aging batteries, it is broken or in severe cases, an explosion accident is occurring. Therefore, research on the stability and life of batteries continues. However, prediction of battery SoH is difficult due to various variables. Data-based artificial intelligence prediction can be made to solve this problem. This paper analyzed the battery data set provided by NASA to predict the remaining life of a lithium-ion battery, extracted the life characteristics, and predicted the SoH through artificial intelligence technology. Support Vector Machine (SVM) and Long Short-Terms Memory (LSTM) were used as artificial intelligence algorithms. As a result, for NASA battery data with temporal mechanism, 3 characteristics were extracted for each data set, and the RMSE of SVM showed lower results than LSTM, showing relatively high accuracy.