BIG DATA DRIVEN LITHIUM-ION BATTERY MODELING METHOD: A DEEP TRANSFER LEARNING APPROACH

Shuangqi Li; Hongwen He; Jianwei Li

doi:10.46855/energy-proceedings-2345

BIG DATA DRIVEN LITHIUM-ION BATTERY MODELING METHOD: A DEEP TRANSFER LEARNING APPROACH

Shuangqi Li, Hongwen He^*, Jianwei Li

^*Corresponding author for this work

School of Mechanical Engineering

Beijing Institute of Technology

Research output: Contribution to journal › Conference article › peer-review

Abstract

Battery is the bottleneck technology of electric vehicles (EVs), which has complex and hardly observable inside chemical reactions. To reduce the training data volume requirement in artificial intelligent algorithm based battery model, this paper presents a deep transfer learning algorithm based battery modeling method. The Deep Belief Network-Extreme Learning Machine (DBN-ELM) algorithm is used for battery modeling issue in this paper to excavate the hidden features in battery data set and improve the accuracy and stability. The results show that the proposed transfer learning algorithm based battery modeling method is able to achieve a highly accurate simulation for battery dynamic characteristics under an insufficient data set, and the mean absolute percentage error of the established model is within 3%.

Original language	English
Journal	Energy Proceedings
Volume	3
DOIs	https://doi.org/10.46855/energy-proceedings-2345
Publication status	Published - 2019
Event	11th International Conference on Applied Energy, ICAE 2019 - Västerås, Sweden Duration: 12 Aug 2019 → 15 Aug 2019

Keywords

deep learning
electric vehicle
lithium-ion battery
modeling
transfer learning

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.46855/energy-proceedings-2345

Cite this

@article{a8b8e9b11ce64cba88e1434330dce50a,

title = "BIG DATA DRIVEN LITHIUM-ION BATTERY MODELING METHOD: A DEEP TRANSFER LEARNING APPROACH",

abstract = "Battery is the bottleneck technology of electric vehicles (EVs), which has complex and hardly observable inside chemical reactions. To reduce the training data volume requirement in artificial intelligent algorithm based battery model, this paper presents a deep transfer learning algorithm based battery modeling method. The Deep Belief Network-Extreme Learning Machine (DBN-ELM) algorithm is used for battery modeling issue in this paper to excavate the hidden features in battery data set and improve the accuracy and stability. The results show that the proposed transfer learning algorithm based battery modeling method is able to achieve a highly accurate simulation for battery dynamic characteristics under an insufficient data set, and the mean absolute percentage error of the established model is within 3%.",

keywords = "deep learning, electric vehicle, lithium-ion battery, modeling, transfer learning",

author = "Shuangqi Li and Hongwen He and Jianwei Li",

note = "Publisher Copyright: {\textcopyright} 2019 ICAE.; 11th International Conference on Applied Energy, ICAE 2019 ; Conference date: 12-08-2019 Through 15-08-2019",

year = "2019",

doi = "10.46855/energy-proceedings-2345",

language = "English",

volume = "3",

journal = "Energy Proceedings",

issn = "2004-2965",

publisher = "Scanditale AB",

}

TY - JOUR

T1 - BIG DATA DRIVEN LITHIUM-ION BATTERY MODELING METHOD

T2 - 11th International Conference on Applied Energy, ICAE 2019

AU - Li, Shuangqi

AU - He, Hongwen

AU - Li, Jianwei

PY - 2019

Y1 - 2019

N2 - Battery is the bottleneck technology of electric vehicles (EVs), which has complex and hardly observable inside chemical reactions. To reduce the training data volume requirement in artificial intelligent algorithm based battery model, this paper presents a deep transfer learning algorithm based battery modeling method. The Deep Belief Network-Extreme Learning Machine (DBN-ELM) algorithm is used for battery modeling issue in this paper to excavate the hidden features in battery data set and improve the accuracy and stability. The results show that the proposed transfer learning algorithm based battery modeling method is able to achieve a highly accurate simulation for battery dynamic characteristics under an insufficient data set, and the mean absolute percentage error of the established model is within 3%.

AB - Battery is the bottleneck technology of electric vehicles (EVs), which has complex and hardly observable inside chemical reactions. To reduce the training data volume requirement in artificial intelligent algorithm based battery model, this paper presents a deep transfer learning algorithm based battery modeling method. The Deep Belief Network-Extreme Learning Machine (DBN-ELM) algorithm is used for battery modeling issue in this paper to excavate the hidden features in battery data set and improve the accuracy and stability. The results show that the proposed transfer learning algorithm based battery modeling method is able to achieve a highly accurate simulation for battery dynamic characteristics under an insufficient data set, and the mean absolute percentage error of the established model is within 3%.

KW - deep learning

KW - electric vehicle

KW - lithium-ion battery

KW - modeling

KW - transfer learning

UR - http://www.scopus.com/inward/record.url?scp=85202592027&partnerID=8YFLogxK

U2 - 10.46855/energy-proceedings-2345

DO - 10.46855/energy-proceedings-2345

M3 - Conference article

AN - SCOPUS:85202592027

SN - 2004-2965

VL - 3

JO - Energy Proceedings

JF - Energy Proceedings

Y2 - 12 August 2019 through 15 August 2019

ER -

BIG DATA DRIVEN LITHIUM-ION BATTERY MODELING METHOD: A DEEP TRANSFER LEARNING APPROACH

Abstract

Keywords

UN SDGs

Access to Document

Other files and links

Fingerprint

Cite this