Using Machine Learning to Predict Charge Depletion Patterns in Electric Vehicles

SAE-PP-00217

1/6/2022

Authors
Abstract
Content
Electric Vehicles are poised to become a major force in worldwide transportation. The heart of an EV, its lithium battery, is what powers the entire vehicle. Unlike gas cars, which will show a vague scale of how much gas is remaining to power a car, electric vehicles generally show a battery percentage, much like in a mobile device. As more of the worldwide population starts to use electric vehicles, the state of the battery must be assessed and projected for any scenario. In this paper, we leverage machine learning algorithms to predict the battery State of Charge. We use a publicly available dataset describing vehicle parameters and trip details for 70 trips in EV BMW™ i3 (60 ah vehicles and evaluate machine learning algorithms for predicting State of Charge Percentage. We utilized a data processing technique (deltas and staggers to extract different perspectives from each trip record and demonstrate that machine learning techniques can be effectively used to predict battery state of charge for a wide range of driving conditions and trip parameters. Finally, we also demonstrated which parameters most influence the State of Charge Prediction.
Meta TagsDetails
DOI
https://doi.org/10.47953/SAE-PP-00217
Citation
Puri, S., "Using Machine Learning to Predict Charge Depletion Patterns in Electric Vehicles," SAE MobilityRxiv™ Preprint, submitted January 6, 2022, https://doi.org/10.47953/SAE-PP-00217.
Additional Details
Publisher
Published
1/6/2022
Product Code
SAE-PP-00217
Content Type
Pre-Print Article
Language
English