Comparison of two probabilistic fatigue damage assessment approaches using prognostic performance metrics

Xuefei Guan, Yongming Liu, Ratneshwar Jha, Abhinav Saxena, Jose Celaya, Kai Geobel

Research output: Contribution to journalArticlepeer-review

23 Scopus citations


In this paper, two probabilistic prognosis updating schemes are compared. One is based on the classical Bayesian approach and the other is based on newly developed maximum relative entropy (MRE) approach. The algorithm performance of the two models is evaluated using a set of recently developed prognostics-based metrics. Various uncertainties from measurements, modeling, and parameter estimations are integrated into the prognosis framework as random input variables for fatigue damage of materials. Measures of response variables are then used to update the statistical distributions of random variables and the prognosis results are updated using posterior distributions. Markov Chain Monte Carlo (MCMC) technique is employed to provide the posterior samples for model updating in the framework. Experimental data are used to demonstrate the operation of the proposed probabilistic prognosis methodology. A set of prognostics-based metrics are employed to quantitatively evaluate the prognosis performance and compare the proposed entropy method with the classical Bayesian updating algorithm. In particular, model accuracy, precision, robustness and convergence are rigorously evaluated in addition to the qualitative visual comparison. Following this, potential development and improvement for the prognostics-based metrics are discussed in detail.

Original languageEnglish (US)
JournalInternational Journal of Prognostics and Health Management
Issue number1
StatePublished - 2011
Externally publishedYes

ASJC Scopus subject areas

  • Computer Science (miscellaneous)
  • Civil and Structural Engineering
  • Safety, Risk, Reliability and Quality
  • Energy Engineering and Power Technology
  • Mechanical Engineering


Dive into the research topics of 'Comparison of two probabilistic fatigue damage assessment approaches using prognostic performance metrics'. Together they form a unique fingerprint.

Cite this