A Note on Shock Model Justification for IFR Distributions
We consider the model in which the failure rate for a device changes when the device is subjected to shocks which occur stochastically over time. We show that increasing failure rate distributions can be obtained by making simple models for the effects of shocks. The results provide a physical motivation for using the Weibull distributions for failure time data. Random failure rates used in Bayesian inference are also obtained in a similar manner by modeling the effects of shocks to be stochastic.
Key WordsFailure rate Poisson process shock model stochastic failure rate Weibull distribution weak convergence
Unable to display preview. Download preview PDF.
- Dykstra, R. L., Laud, P.W. (1980). A Bayesian nonparametric approach toward reliability. Annals of Statistics (to appear).Google Scholar