Lithium-ion battery remaining useful life prediction with long short-term memory recurrent neural network

Yuefeng Liu; Guangquan Zhao; Xiyuan Peng; Cong Hu

Lithium-ion battery remaining useful life prediction with long short-term memory recurrent neural network

Yuefeng Liu
, Guangquan Zhao
, Xiyuan Peng
, Cong Hu

School of Electronics and Information Engineering

Harbin Institute of Technology
Inner Mongolia University of Science and Technology
Guilin University of Electronic Technology

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

21 Scopus citations

Abstract

Lithium-ion batteries play critical roles in many electronicdevices. It is necessary to develop a reliable and accurateremaining useful life (RUL) prediction approach to providetimely maintenance or replacement of battery systems. Anovel RUL prediction approach based on Long Short-termMemory (LSTM) Recurrent Neural Network (RNN) isproposed in this paper. LSTM is able to capture long-termdependencies and model sequential data among the capacitydegradation of lithium-ion batteries. The advantages of ourproposed method include: 1) obtaining high predictionaccuracy without accurate physics-based model or expertiseand 2) decreasing the cumulation errors by multi-step aheadprediction each time, while traditional RUL method predictsone-step ahead once and then uses the current estimatedvalue to predict next one, which causes cumulation errorsincreased. The Center for Advanced Life Cycle Engineering(CALCE) battery datasets are used to demonstrate theeffectiveness of the proposed method. The results show that,compared with echo state networks (ESN), the proposedmethod has higher accuracy, more stable and reliableperformance for lithium-ion batteries RUL prediction.

Original language	English
Title of host publication	PHM 2017 - Proceedings of the Annual Conference of the Prognostics and Health Management Society 2017
Editors	Matthew J. Daigle, Anibal Bregon
Publisher	Prognostics and Health Management Society
Pages	510-516
Number of pages	7
ISBN (Electronic)	9781936263059
State	Published - 2017
Event	9th Annual Conference of the Prognostics and Health Management Society, PHM 2017 - St. Petersburg, United States Duration: 2 Oct 2017 → 5 Oct 2017

Publication series

Name	Proceedings of the Annual Conference of the Prognostics and Health Management Society, PHM
ISSN (Print)	2325-0178

Conference

Conference	9th Annual Conference of the Prognostics and Health Management Society, PHM 2017
Country/Territory	United States
City	St. Petersburg
Period	2/10/17 → 5/10/17

UN SDGs

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

SDG 7 Affordable and Clean Energy

Cite this

Liu, Y., Zhao, G., Peng, X., & Hu, C. (2017). Lithium-ion battery remaining useful life prediction with long short-term memory recurrent neural network. In M. J. Daigle, & A. Bregon (Eds.), PHM 2017 - Proceedings of the Annual Conference of the Prognostics and Health Management Society 2017 (pp. 510-516). (Proceedings of the Annual Conference of the Prognostics and Health Management Society, PHM). Prognostics and Health Management Society.

Liu, Yuefeng ; Zhao, Guangquan ; Peng, Xiyuan et al. / Lithium-ion battery remaining useful life prediction with long short-term memory recurrent neural network. PHM 2017 - Proceedings of the Annual Conference of the Prognostics and Health Management Society 2017. editor / Matthew J. Daigle ; Anibal Bregon. Prognostics and Health Management Society, 2017. pp. 510-516 (Proceedings of the Annual Conference of the Prognostics and Health Management Society, PHM).

@inproceedings{67e9a53fdfcd4a5dbd8d7f85c6a40fcb,

title = "Lithium-ion battery remaining useful life prediction with long short-term memory recurrent neural network",

abstract = "Lithium-ion batteries play critical roles in many electronicdevices. It is necessary to develop a reliable and accurateremaining useful life (RUL) prediction approach to providetimely maintenance or replacement of battery systems. Anovel RUL prediction approach based on Long Short-termMemory (LSTM) Recurrent Neural Network (RNN) isproposed in this paper. LSTM is able to capture long-termdependencies and model sequential data among the capacitydegradation of lithium-ion batteries. The advantages of ourproposed method include: 1) obtaining high predictionaccuracy without accurate physics-based model or expertiseand 2) decreasing the cumulation errors by multi-step aheadprediction each time, while traditional RUL method predictsone-step ahead once and then uses the current estimatedvalue to predict next one, which causes cumulation errorsincreased. The Center for Advanced Life Cycle Engineering(CALCE) battery datasets are used to demonstrate theeffectiveness of the proposed method. The results show that,compared with echo state networks (ESN), the proposedmethod has higher accuracy, more stable and reliableperformance for lithium-ion batteries RUL prediction.",

author = "Yuefeng Liu and Guangquan Zhao and Xiyuan Peng and Cong Hu",

note = "Publisher Copyright: {\textcopyright} 2017 Prognostics and Health Management Society. All rights reserved.; 9th Annual Conference of the Prognostics and Health Management Society, PHM 2017 ; Conference date: 02-10-2017 Through 05-10-2017",

year = "2017",

language = "英语",

series = "Proceedings of the Annual Conference of the Prognostics and Health Management Society, PHM",

publisher = "Prognostics and Health Management Society",

pages = "510--516",

editor = "Daigle, \{Matthew J.\} and Anibal Bregon",

booktitle = "PHM 2017 - Proceedings of the Annual Conference of the Prognostics and Health Management Society 2017",

address = "美国",

}

Liu, Y, Zhao, G , Peng, X & Hu, C 2017, Lithium-ion battery remaining useful life prediction with long short-term memory recurrent neural network. in MJ Daigle & A Bregon (eds), PHM 2017 - Proceedings of the Annual Conference of the Prognostics and Health Management Society 2017. Proceedings of the Annual Conference of the Prognostics and Health Management Society, PHM, Prognostics and Health Management Society, pp. 510-516, 9th Annual Conference of the Prognostics and Health Management Society, PHM 2017, St. Petersburg, United States, 2/10/17.

Lithium-ion battery remaining useful life prediction with long short-term memory recurrent neural network. / Liu, Yuefeng; Zhao, Guangquan ; Peng, Xiyuan et al.
PHM 2017 - Proceedings of the Annual Conference of the Prognostics and Health Management Society 2017. ed. / Matthew J. Daigle; Anibal Bregon. Prognostics and Health Management Society, 2017. p. 510-516 (Proceedings of the Annual Conference of the Prognostics and Health Management Society, PHM).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Lithium-ion battery remaining useful life prediction with long short-term memory recurrent neural network

AU - Liu, Yuefeng

AU - Zhao, Guangquan

AU - Peng, Xiyuan

AU - Hu, Cong

PY - 2017

Y1 - 2017

N2 - Lithium-ion batteries play critical roles in many electronicdevices. It is necessary to develop a reliable and accurateremaining useful life (RUL) prediction approach to providetimely maintenance or replacement of battery systems. Anovel RUL prediction approach based on Long Short-termMemory (LSTM) Recurrent Neural Network (RNN) isproposed in this paper. LSTM is able to capture long-termdependencies and model sequential data among the capacitydegradation of lithium-ion batteries. The advantages of ourproposed method include: 1) obtaining high predictionaccuracy without accurate physics-based model or expertiseand 2) decreasing the cumulation errors by multi-step aheadprediction each time, while traditional RUL method predictsone-step ahead once and then uses the current estimatedvalue to predict next one, which causes cumulation errorsincreased. The Center for Advanced Life Cycle Engineering(CALCE) battery datasets are used to demonstrate theeffectiveness of the proposed method. The results show that,compared with echo state networks (ESN), the proposedmethod has higher accuracy, more stable and reliableperformance for lithium-ion batteries RUL prediction.

AB - Lithium-ion batteries play critical roles in many electronicdevices. It is necessary to develop a reliable and accurateremaining useful life (RUL) prediction approach to providetimely maintenance or replacement of battery systems. Anovel RUL prediction approach based on Long Short-termMemory (LSTM) Recurrent Neural Network (RNN) isproposed in this paper. LSTM is able to capture long-termdependencies and model sequential data among the capacitydegradation of lithium-ion batteries. The advantages of ourproposed method include: 1) obtaining high predictionaccuracy without accurate physics-based model or expertiseand 2) decreasing the cumulation errors by multi-step aheadprediction each time, while traditional RUL method predictsone-step ahead once and then uses the current estimatedvalue to predict next one, which causes cumulation errorsincreased. The Center for Advanced Life Cycle Engineering(CALCE) battery datasets are used to demonstrate theeffectiveness of the proposed method. The results show that,compared with echo state networks (ESN), the proposedmethod has higher accuracy, more stable and reliableperformance for lithium-ion batteries RUL prediction.

UR - https://www.scopus.com/pages/publications/85071661455

M3 - 会议稿件

AN - SCOPUS:85071661455

T3 - Proceedings of the Annual Conference of the Prognostics and Health Management Society, PHM

SP - 510

EP - 516

BT - PHM 2017 - Proceedings of the Annual Conference of the Prognostics and Health Management Society 2017

A2 - Daigle, Matthew J.

A2 - Bregon, Anibal

PB - Prognostics and Health Management Society

T2 - 9th Annual Conference of the Prognostics and Health Management Society, PHM 2017

Y2 - 2 October 2017 through 5 October 2017

ER -

Liu Y, Zhao G , Peng X, Hu C. Lithium-ion battery remaining useful life prediction with long short-term memory recurrent neural network. In Daigle MJ, Bregon A, editors, PHM 2017 - Proceedings of the Annual Conference of the Prognostics and Health Management Society 2017. Prognostics and Health Management Society. 2017. p. 510-516. (Proceedings of the Annual Conference of the Prognostics and Health Management Society, PHM).

Lithium-ion battery remaining useful life prediction with long short-term memory recurrent neural network

Abstract

Publication series

Conference

UN SDGs

Other files and links

Fingerprint

Cite this