Abstract:Accurately estimating the remaining useful life (RUL) of lithium-ion batteries is crucial for the safe and reliable operation of batteries in electric vehicles. Due to the slow aging process and complex chemical reactions of batteries, it is challenging to obtain the complete battery lifecycle data to predict the RUL. To embrace these challenges, we propose an intelligent Transformer network with the improved self-attention mechanism based on data of the key interval area. Firstly, the knee-point and knee-onset in the capacity decay curve of the battery are identified by the Bacon-Watts model. The feature data that does not meet the minimum aging period between the knee-point and knee-onset in the aging curve is eliminated to ensure the accuracy and quickness of the prediction. Secondly, the denoising autoencoder (DAE) is used to mine the correlation between the aging characteristics in this interval and the low-dimensional hidden feature of battery aging is reconstructed. Finally, the Transformer network with improved multi-head self-attention mechanism is taken to capture the dependencies between features at different positions. A feed-forward neural network is employed to determine the weights of the features at the different sample time, which are then used in conjunction with fully connected layers and a prediction layer to estimate the remaining lifespan of lithium batteries. The proposed method is validated using the CALCE battery dataset. Simulation results show that the method achieves accurate and fast health prognosis with an average error of 0.0091 for the accuracy metric RE when only key feature interval data are used. There is also a significant advantage in computation time compared to other state-of-the-art algorithms, and our method can provide more accurate and faster health prognosis of batteries.

An Interpretable Semi-Supervised Learning Approach for Battery Lifespan Early Prognostic

A Physics-Informed Hybrid Data-Driven Approach with Generative Electrode-Level Features for Lithium-Ion Battery Health Prognostics

A Physics-Informed Hybrid Multitask Learning for Lithium-Ion Battery Full-Life Aging Estimation at Early Lifetime

Battery Early Prognostics Based on Pseudo Meta-Learning

A deep learning approach incorporating attention mechanism and transfer learning for lithium-ion battery lifespan prediction

Early prediction of lithium-ion battery lifetime via a hybrid deep learning model

Enhanced battery life prediction with reduced data demand via semi-supervised representation learning

A novel denoising autoencoder hybrid network for remaining useful life estimation of lithium‐ion batteries

Research on the remaining useful life prediction method for lithium-ion batteries by fusion of feature engineering and deep learning

Battery prognostics and health management from a machine learning perspective

A novel hybrid model for lithium-ion batteries lifespan prediction with high accuracy and interpretability

Flexible health prognosis of battery nonlinear aging using temporal transfer learning

An end-to-end neural network framework for state-of-health estimation and remaining useful life prediction of electric vehicle lithium batteries

Semi-supervised learning for explainable few-shot battery lifetime prediction

An Empirical-informed Model for the Early Degradation Trajectory Prediction of Lithium-ion Battery

A Battery Prognostics and Health Management Technique Based on Knee Critical Interval and Linear Complexity Self-Attention Transformer in Electric Vehicles

Lithium-ion battery degradation trajectory early prediction with synthetic dataset and deep learning

Battery health management using physics-informed machine learning: Online degradation modeling and remaining useful life prediction

Lithium battery state-of-health estimation and remaining useful lifetime prediction based on non-parametric aging model and particle filter algorithm

Health and lifespan prediction considering degradation patterns of lithium-ion batteries based on transferable attention neural network

Improving in-situ life prediction and classification performance by capturing both the present state and evolution rate of battery aging