Mis-specification effect analyses on the estimated mean-time-to-failure of inverse Gaussian and Wiener degradation process

Abstract

In recent years degradation analysis has widely used to access lifetime of highly reliable products If there exists quality characteristics whose degradation paths can be related to the reliability of the product then collecting degradation data can help with evaluating product reliability and lifetime In this paper motivated by laser data we investigate the mis-specification effect on the prediction of product’s MTTF (mean-time-to-failure) when the wrong degradation model is fitted to data On the basis of the maximum likelihood under a wrong model theory we first derive mean and variance of the product’s MTTF when the true degradation model comes from inverse Gaussian process but is wrongly assumed to be Wiener process In addition laser data are used to illustrate the model mis-specification effect A simulation study is carried out to evaluate the bias of the model mis-specification using which we show that the simulation results are quite close to the theoretical ones The result reveals that the effect on the accuracy of the product’s MTTF prediction depends on the value of critical value and the parameter of the inverse Gaussian process In addition the effects on the precision of the product’s MTTF prediction are influenced by the parameters of the inverse Gaussian process

Date of Award	2019
Original language	English
Supervisor	Cheng-Hung Hu (Supervisor)