Improving Lithium-ion Battery Life Prediction with an Enhanced Multi-Layer Perceptron Model Featuring Autoencoder-Based Feature Extraction

This study proposes an improved multi-layer perceptron (MLP) model based on autoencoder(AE) feature extraction to predict the life of lithium-ion batteries. In this study, the dimensions of the battery status data were firstly reduced by the autoencoder, and then the extracted features were used to train the MLP model to improve the prediction accuracy of the remaining useful life (RUL) of the battery. In addition, this study also discussed the changes of battery performance at different charging rates, and verified the superiority of the proposed AE-MLP model in predicting the life of lithium batteries at different charging rates by comparing with the prior art. The results show that the AE-MLP model is superior to the traditional MLP model in terms of prediction accuracy and robustness.