Enhancing Battery Degradation Prediction using Recurrent and Convolutional Neural Networks

Abstract

In the field of battery technology, the ability to predict degradation accurately is of paramount importance.This study aims to develop and assess a data-driven prognostics framework for forecasting capacity and power fade in lithium-ion batteries. Leveraging machine learning techniques, specifically recurrent neural networks (RNNs) and convolutional neural networks (CNNs), this framework models the intricate relationships between battery performance and aging effects. Through the utilization of historical battery data, it facilitates accurate predictions of future degradation trends. The performance of the models is meticulously evaluated using relevant metrics, including the Root Mean Squared Error (RMSE) and Mean absolute error (MAE). Furthermore, the paper provides a comparative analysis to gauge the accuracy and efficacy of RNNs and CNNs models for battery prognostics, contributing to a deeper understanding of their respective capabilities. It is demonstrated that RNNs excel in predicting both resistance and capacitance degradation. These findings have the potential to significantly enhance the precision and efficiency of battery degradation predictions, with broad applications across industries.