An integrated probabilistic approach to lithium-ion battery remaining useful life estimation

Datong Liu, Wei Xie, Haitao Liao, Yu Peng

Research output: Contribution to journalArticle

52 Citations (Scopus)

Abstract

Estimating lithium-ion battery remaining useful life (RUL) is a key issue in an intelligent battery management system. This paper presents an integrated prognostic approach that unifies two types of health indices (HIs), battery capacity and time interval of equal discharging voltage difference series, to perform direct and indirect RUL estimation for lithium-ion battery. To satisfy different practical requirements, a data-driven monotonic echo state networks (MONESNs) algorithm is adopted to track the nonlinear patterns of battery degradation. The main contributions of this paper are: 1) to enhance the predictive capability of each HI and identify its failure threshold by implementing an HI correlation model and cycle life threshold transformation and 2) to increase the computational stability of the proposed approach through the ensemble of MONESN submodels that can also describe the prognostic uncertainty. Essentially, this approach constitutes a probabilistic integration and data-driven prognostic framework with uncertainty management capability. Two sets of industrial lithium-ion battery data are used to show the capability of the proposed approach. It is expected that this approach can be broadly applied to other application areas, where data-driven prognostic approaches are needed.

Original languageEnglish (US)
Article number6895159
Pages (from-to)660-670
Number of pages11
JournalIEEE Transactions on Instrumentation and Measurement
Volume64
Issue number3
DOIs
StatePublished - Mar 1 2015

Fingerprint

electric batteries
lithium
Health
health
ions
Life cycle
echoes
Degradation
thresholds
management systems
Electric potential
Lithium-ion batteries
estimating
degradation
intervals
requirements
cycles
Uncertainty
electric potential
Battery management systems

Keywords

  • Ensemble learning (EL)
  • lithium-ion battery
  • maximum likelihood estimation (MLE)
  • prognostic uncertainty
  • remaining useful life (RUL)

ASJC Scopus subject areas

  • Instrumentation
  • Electrical and Electronic Engineering

Cite this

An integrated probabilistic approach to lithium-ion battery remaining useful life estimation. / Liu, Datong; Xie, Wei; Liao, Haitao; Peng, Yu.

In: IEEE Transactions on Instrumentation and Measurement, Vol. 64, No. 3, 6895159, 01.03.2015, p. 660-670.

Research output: Contribution to journalArticle

@article{c1c6016418394a3984189e569b87594a,
title = "An integrated probabilistic approach to lithium-ion battery remaining useful life estimation",
abstract = "Estimating lithium-ion battery remaining useful life (RUL) is a key issue in an intelligent battery management system. This paper presents an integrated prognostic approach that unifies two types of health indices (HIs), battery capacity and time interval of equal discharging voltage difference series, to perform direct and indirect RUL estimation for lithium-ion battery. To satisfy different practical requirements, a data-driven monotonic echo state networks (MONESNs) algorithm is adopted to track the nonlinear patterns of battery degradation. The main contributions of this paper are: 1) to enhance the predictive capability of each HI and identify its failure threshold by implementing an HI correlation model and cycle life threshold transformation and 2) to increase the computational stability of the proposed approach through the ensemble of MONESN submodels that can also describe the prognostic uncertainty. Essentially, this approach constitutes a probabilistic integration and data-driven prognostic framework with uncertainty management capability. Two sets of industrial lithium-ion battery data are used to show the capability of the proposed approach. It is expected that this approach can be broadly applied to other application areas, where data-driven prognostic approaches are needed.",
keywords = "Ensemble learning (EL), lithium-ion battery, maximum likelihood estimation (MLE), prognostic uncertainty, remaining useful life (RUL)",
author = "Datong Liu and Wei Xie and Haitao Liao and Yu Peng",
year = "2015",
month = "3",
day = "1",
doi = "10.1109/TIM.2014.2348613",
language = "English (US)",
volume = "64",
pages = "660--670",
journal = "IEEE Transactions on Instrumentation and Measurement",
issn = "0018-9456",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "3",

}

TY - JOUR

T1 - An integrated probabilistic approach to lithium-ion battery remaining useful life estimation

AU - Liu, Datong

AU - Xie, Wei

AU - Liao, Haitao

AU - Peng, Yu

PY - 2015/3/1

Y1 - 2015/3/1

N2 - Estimating lithium-ion battery remaining useful life (RUL) is a key issue in an intelligent battery management system. This paper presents an integrated prognostic approach that unifies two types of health indices (HIs), battery capacity and time interval of equal discharging voltage difference series, to perform direct and indirect RUL estimation for lithium-ion battery. To satisfy different practical requirements, a data-driven monotonic echo state networks (MONESNs) algorithm is adopted to track the nonlinear patterns of battery degradation. The main contributions of this paper are: 1) to enhance the predictive capability of each HI and identify its failure threshold by implementing an HI correlation model and cycle life threshold transformation and 2) to increase the computational stability of the proposed approach through the ensemble of MONESN submodels that can also describe the prognostic uncertainty. Essentially, this approach constitutes a probabilistic integration and data-driven prognostic framework with uncertainty management capability. Two sets of industrial lithium-ion battery data are used to show the capability of the proposed approach. It is expected that this approach can be broadly applied to other application areas, where data-driven prognostic approaches are needed.

AB - Estimating lithium-ion battery remaining useful life (RUL) is a key issue in an intelligent battery management system. This paper presents an integrated prognostic approach that unifies two types of health indices (HIs), battery capacity and time interval of equal discharging voltage difference series, to perform direct and indirect RUL estimation for lithium-ion battery. To satisfy different practical requirements, a data-driven monotonic echo state networks (MONESNs) algorithm is adopted to track the nonlinear patterns of battery degradation. The main contributions of this paper are: 1) to enhance the predictive capability of each HI and identify its failure threshold by implementing an HI correlation model and cycle life threshold transformation and 2) to increase the computational stability of the proposed approach through the ensemble of MONESN submodels that can also describe the prognostic uncertainty. Essentially, this approach constitutes a probabilistic integration and data-driven prognostic framework with uncertainty management capability. Two sets of industrial lithium-ion battery data are used to show the capability of the proposed approach. It is expected that this approach can be broadly applied to other application areas, where data-driven prognostic approaches are needed.

KW - Ensemble learning (EL)

KW - lithium-ion battery

KW - maximum likelihood estimation (MLE)

KW - prognostic uncertainty

KW - remaining useful life (RUL)

UR - http://www.scopus.com/inward/record.url?scp=85027951825&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85027951825&partnerID=8YFLogxK

U2 - 10.1109/TIM.2014.2348613

DO - 10.1109/TIM.2014.2348613

M3 - Article

VL - 64

SP - 660

EP - 670

JO - IEEE Transactions on Instrumentation and Measurement

JF - IEEE Transactions on Instrumentation and Measurement

SN - 0018-9456

IS - 3

M1 - 6895159

ER -