Monitoring State of Health and State of Charge of Lithium-Ion Batteries Using Machine Learning Techniques

Abstract

Lithium-ion batteries are used in a wide range of applications. However, monitoring these batteries effectively is a challenge. There have been several attempts to efficiently estimate the battery state by fitting semi-empirical models. However, these methods tend to be computationally costly. This paper aims to solve this problem using a battery monitoring prototype based on supervised machine learning by estimating the battery’s health and the charge contained by it It can be efficiently designed by making a few changes in the production cycle. The dataset obtained by the manufacturer can be used to train a machine learning model. These models then can be used to estimate the behavioral patterns of the battery in real-time which gives the user an idea about the performance of the battery at any instance. Further, this monitoring system can be extended on a large scale with the help of Internet of Things as thousands of batteries can be monitored using a single server running all the algorithms, thus reducing cost for large-scale applications.