Prediction of SOH and RUL for Li-Ion Batteries in EV Based on AttentiveLSTM Multi-Task Model

This study introduces innovative approach for predicting the State of Health (SOH) and Remaining Useful Life (RUL) of lithium-ion batteries, leveraging datasets from NASA and Oxford to establish a robust predictive framework. A key highlight is the development of a novel neural network architecture defined as AttentiveLSTM which integrates Long Short-Term Memory (LSTM) networks with Transformer mechanisms to enhance feature extraction and time-series forecasting. The research addresses critical challenges in the domain, including the non-linear behavior of battery degradation, diverse operating conditions, and the scarcity of historical data. The proposed AttentiveLSTM model surpasses existing approaches in predictive accuracy for both SOH and RUL. Additionally, it introduces an advanced objective function combining Denoising Autoencoder (DAE) loss functions with prediction loss to improve model performance. This work not only advances predictive modeling techniques but also contributes to the broader goal of enabling more efficient and sustainable use of electric vehicle (EV) batteries, thereby supporting the transition to eco-friendly industrial transportation systems.