Remaining Useful Life Assessment for Lithium-Ion Batteries Using CNN-LSTM-DNN Hybrid Method

Abstract

The prediction of a Lithium-ion battery's lifetime is very important for ensuring safety and reliability. In addition, it is utilized as an early warning system to prevent the battery's failure. Recent advance in Machine Learning (ML) is an enabler for new data-driven estimation approaches. In this paper, we suggest a hybrid method, named the CNN-LSTM-DNN, which is a combination of Convolutional Neural Network (CNN), Long Short Term Memory (LSTM), and Deep Neural Networks (DNN), for the estimation of the battery's remaining useful life (RUL) and improving prediction accuracy with acceptable execution time. A comparison against various ML estimation algorithms is carried out to show the superiority of the proposed hybrid estimation approach. For that, three statistical indicators, i.e., the MAE, R <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> , and RMSE, are selected to assess numerically the prediction results. Experimental validation is performed using two datasets of different lithium-ion batteries from NASA and CALCE. Thus, results reveal that hybrid methods perform better than the single ones, also the effectiveness of the suggested method in reducing the prediction error and in achieving better RUL prediction performance compared to the other methods.