A machine learning-based framework for online prediction of battery ageing trajectory and lifetime using histogram data

Show moreAdd to Mendeleyhttps://doi.org/10.1016/j.jpowsour.2022.231110Get rights and content

Open access

Highlightso

A comprehensive feature pool based on the battery usage histogram data is formed.

A systematic machine learning framework based on histogram data is introduced.

A comparison of the four most commonly used algorithms on three datasets is shown.

An individualisation process is developed during the online deployment.

A real-world PHEV fleet is used to verify the efficacy of the proposed framework.

Abstract

Accurately predicting batteries' ageing trajectory and remaining useful life is not only required to ensure safe and reliable operation of electric vehicles (EVs) but is also the fundamental step towards health-conscious use and residual value assessment of the battery. The non-linearity, wide range of operating conditions, and cell to cell variations make battery health prediction challenging. This paper proposes a prediction framework that is based on a combination of global models offline developed by different machine learning methods and cell individualised models that are online adapted. For any format of raw data collected under diverse operating conditions, statistic properties of histograms can be still extracted and used as features to learn battery ageing. Our framework is trained and tested on three large datasets, one being retrieved from 7296 plug-in hybrid EVs. While the best global models achieve 0.93% mean absolute percentage error (MAPE) on laboratory data and 1.41% MAPE on the real-world fleet data, the adaptation algorithm further reduced the errors by up to 13.7%, all requiring low computational power and memory. Overall, this work proves the feasibility and benefits of using histogram data and also highlights how online adaptation can be used to improve predictions.

Graphical abstract

Keywords

Lithium-ion batteries

State of health prediction

Remaining useful life

Machine learning

Online adaptive learning

Real-world fleet data