Genetic identification and fisher identifiability analysis of the Doyle-Fuller-Newman model from experimental cycling of a LiFePO 4 cell

Joel C. Forman, Scott J. Moura, Jeffrey L. Stein, Hosam Kadry Fathy

Research output: Contribution to journalArticle

112 Citations (Scopus)

Abstract

This article examines the problem of identifying the physical parameters of fundamental electrochemistry-based battery models from non-invasive voltage/current cycling tests. The article is particularly motivated by the problem of fitting the Doyle-Fuller-Newman (DFN) battery model to lithium-ion battery cycling data. Previous research in the literature identifies subsets of the DFN model's parameter experimentally. In contrast, this article makes the two unique contributions of: (i) identifying the full set of DFN model parameters from cycling data using a genetic algorithm (GA), and (ii) assessing the accuracy and identifiability of the resulting full parameter set using Fisher information. The specific battery used within this study has lithium iron phosphate cathode chemistry and is intended for high-power applications such as plug-in hybrid electric vehicles (PHEVs). We use seven experimental cycling data sets for model fitting and validation, six of them derived from PHEV drive cycles. This makes the identified parameter values appropriate for PHEV battery simulation and model-based design and control optimization.

Original languageEnglish (US)
Pages (from-to)263-275
Number of pages13
JournalJournal of Power Sources
Volume210
DOIs
StatePublished - Jul 15 2012

Fingerprint

Plug-in hybrid vehicles
electric batteries
electric hybrid vehicles
cycles
plugs
cells
lithium
Fisher information
design optimization
Electrochemistry
electrochemistry
Lithium
genetic algorithms
set theory
phosphates
Phosphates
Cathodes
Iron
Genetic algorithms
cathodes

All Science Journal Classification (ASJC) codes

  • Renewable Energy, Sustainability and the Environment
  • Energy Engineering and Power Technology
  • Physical and Theoretical Chemistry
  • Electrical and Electronic Engineering

Cite this

Forman, Joel C. ; Moura, Scott J. ; Stein, Jeffrey L. ; Fathy, Hosam Kadry. / Genetic identification and fisher identifiability analysis of the Doyle-Fuller-Newman model from experimental cycling of a LiFePO 4 cell. In: Journal of Power Sources. 2012 ; Vol. 210. pp. 263-275.
@article{21ed19759043486985bbc73de517b07d,
title = "Genetic identification and fisher identifiability analysis of the Doyle-Fuller-Newman model from experimental cycling of a LiFePO 4 cell",
abstract = "This article examines the problem of identifying the physical parameters of fundamental electrochemistry-based battery models from non-invasive voltage/current cycling tests. The article is particularly motivated by the problem of fitting the Doyle-Fuller-Newman (DFN) battery model to lithium-ion battery cycling data. Previous research in the literature identifies subsets of the DFN model's parameter experimentally. In contrast, this article makes the two unique contributions of: (i) identifying the full set of DFN model parameters from cycling data using a genetic algorithm (GA), and (ii) assessing the accuracy and identifiability of the resulting full parameter set using Fisher information. The specific battery used within this study has lithium iron phosphate cathode chemistry and is intended for high-power applications such as plug-in hybrid electric vehicles (PHEVs). We use seven experimental cycling data sets for model fitting and validation, six of them derived from PHEV drive cycles. This makes the identified parameter values appropriate for PHEV battery simulation and model-based design and control optimization.",
author = "Forman, {Joel C.} and Moura, {Scott J.} and Stein, {Jeffrey L.} and Fathy, {Hosam Kadry}",
year = "2012",
month = "7",
day = "15",
doi = "10.1016/j.jpowsour.2012.03.009",
language = "English (US)",
volume = "210",
pages = "263--275",
journal = "Journal of Power Sources",
issn = "0378-7753",
publisher = "Elsevier",

}

Genetic identification and fisher identifiability analysis of the Doyle-Fuller-Newman model from experimental cycling of a LiFePO 4 cell. / Forman, Joel C.; Moura, Scott J.; Stein, Jeffrey L.; Fathy, Hosam Kadry.

In: Journal of Power Sources, Vol. 210, 15.07.2012, p. 263-275.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Genetic identification and fisher identifiability analysis of the Doyle-Fuller-Newman model from experimental cycling of a LiFePO 4 cell

AU - Forman, Joel C.

AU - Moura, Scott J.

AU - Stein, Jeffrey L.

AU - Fathy, Hosam Kadry

PY - 2012/7/15

Y1 - 2012/7/15

N2 - This article examines the problem of identifying the physical parameters of fundamental electrochemistry-based battery models from non-invasive voltage/current cycling tests. The article is particularly motivated by the problem of fitting the Doyle-Fuller-Newman (DFN) battery model to lithium-ion battery cycling data. Previous research in the literature identifies subsets of the DFN model's parameter experimentally. In contrast, this article makes the two unique contributions of: (i) identifying the full set of DFN model parameters from cycling data using a genetic algorithm (GA), and (ii) assessing the accuracy and identifiability of the resulting full parameter set using Fisher information. The specific battery used within this study has lithium iron phosphate cathode chemistry and is intended for high-power applications such as plug-in hybrid electric vehicles (PHEVs). We use seven experimental cycling data sets for model fitting and validation, six of them derived from PHEV drive cycles. This makes the identified parameter values appropriate for PHEV battery simulation and model-based design and control optimization.

AB - This article examines the problem of identifying the physical parameters of fundamental electrochemistry-based battery models from non-invasive voltage/current cycling tests. The article is particularly motivated by the problem of fitting the Doyle-Fuller-Newman (DFN) battery model to lithium-ion battery cycling data. Previous research in the literature identifies subsets of the DFN model's parameter experimentally. In contrast, this article makes the two unique contributions of: (i) identifying the full set of DFN model parameters from cycling data using a genetic algorithm (GA), and (ii) assessing the accuracy and identifiability of the resulting full parameter set using Fisher information. The specific battery used within this study has lithium iron phosphate cathode chemistry and is intended for high-power applications such as plug-in hybrid electric vehicles (PHEVs). We use seven experimental cycling data sets for model fitting and validation, six of them derived from PHEV drive cycles. This makes the identified parameter values appropriate for PHEV battery simulation and model-based design and control optimization.

UR - http://www.scopus.com/inward/record.url?scp=84859627280&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84859627280&partnerID=8YFLogxK

U2 - 10.1016/j.jpowsour.2012.03.009

DO - 10.1016/j.jpowsour.2012.03.009

M3 - Article

AN - SCOPUS:84859627280

VL - 210

SP - 263

EP - 275

JO - Journal of Power Sources

JF - Journal of Power Sources

SN - 0378-7753

ER -