Prognostics Health Estimation of Lithium-ion Batteries in Charge-Decay Estimation Uncertainties – A Comparative Analysis

Chinedu I Ossai

doi:10.36001/ijphm.2018.v9i2.2703

Abstract

This study uses nonlinear mixed effect-based degradation modeling that considers the influence of uncertainties on the state-of-charge of lithium-ion batteries to determine the State-of-Health (SOH) of the batteries at different End-of-Life (EOL) failure thresholds. The results of the analysis obtained with lithium-ion batteries data from NASA Ames Centre repository, confirms that the SOH of the batteries is influenced by the uncertainties. This is because the random effects models show a better correlation with the experimental data than the fixed effects models that have not considered uncertainty. It is important therefore that battery prognosis is done in consideration of these parametric uncertainties, to forestall poor estimation of the SOH of the lithium-ion batteries at various stages of the lifecycle. Seeing that the presence of uncertainties could result in unwarranted failures of assets powered by the batteries, due to over-estimation of the remaining useful life (RUL) or capital loss, due to early decommissioning of efficient batteries when the RUL is under-estimated.

Highlights

One of the challenges in asset integrity management is the ability to estimate the reliability of the facilities, as longer lifecycles of assets resulting from advanced designs and manufacturing technologies make it more difficult to obtain failure results from failure tests analysis
Given the fact that this omission could be a fundamental source of faulty prognosis and unreliable remaining useful life (RUL) estimation, it became imperative that this study explores the influences of uncertainties on the lifetime estimation of lithium-ion battery
This comparative study of the fixed and random effects models, for remaining useful life estimation of the lithium-ion battery was intended to show the influence of uncertainties, which can originate from the manufacturing process, measurements and operational environmental conditions of the battery

Summary

Introduction

One of the challenges in asset integrity management is the ability to estimate the reliability of the facilities, as longer lifecycles of assets resulting from advanced designs and manufacturing technologies make it more difficult to obtain failure results from failure tests analysis. Historical failure records from component life tests, which used to be the primary source of information for reliability estimation is not providing enough information for longtime decisions as failure trend data are becoming sketchier. This longevity of components, though a good thing, has impacted on the validity and accuracy of existing SOH estimation methodologies as rudimentary life data analysis does not provide enough information for the study of Chinedu I. I, have to make use of degradation trends and physical / phenomenological models of degradation as an alternative tool for the prognostics of components, subsystems, and systems of assets

Objectives

Results

Conclusion