Inference on remaining useful life under gamma degradation models with random effects

Man Ho Alpha LING, Hon Keung Tony NG, Kwok-Leung TSUI

Research output: Chapter in Book/Report/Conference proceedingChapters

Abstract

Prognostics and system health management becomes an important topic in modern reliability study. In prognostics and system health management, remaining useful life is one of the vital indexes to yield an advance warning of impending failure in a system, thereby helping in executing preventive actions prior to failure occurrence and helping in making maintenance decisions. To obtain precise statistical inference on the remaining useful life, we consider degradation models that incorporate unit-specific random effects that model heterogeneity in the degradation of distinct systems, and propose a parametric bootstrap confidence interval for the remaining useful life of each system. A Monte Carlo simulation study is carried out to evaluate the performance of the proposed methodology. To illustrate the suggested model and inferential methods, a real data set of light intensity of light emitting diodes is analyzed. Copyright © 2017 Springer Nature Singapore Pte Ltd.
Original languageEnglish
Title of host publicationStatistical modeling for degradation data
EditorsDing-Geng (Din) CHEN, Yuhlong LIO, Hon Keung Tony NG, Tzong-Ru TSAI
Place of PublicationSingapore
PublisherSpringer
Pages253-266
ISBN (Electronic)9789811051944
ISBN (Print)9789811051937
DOIs
Publication statusPublished - 2017

Citation

Ling, M. H., Ng, H. K. T., & Tsui, K.-L. (2017). Inference on remaining useful life under gamma degradation models with random effects. In D.-G. Chen, Y. Lio, H. K. T. Ng, & T.-R. Tsai (Eds.), Statistical modeling for degradation data (pp. 253-266). Singapore: Springer.

Keywords

  • Bootstrap
  • Degradation models
  • Gamma process
  • Random effects
  • Remaining useful life
  • Maximum likelihood estimation

Fingerprint Dive into the research topics of 'Inference on remaining useful life under gamma degradation models with random effects'. Together they form a unique fingerprint.