Modeling Repairable System with Covariates and Interval Failure Data

This thesis presents an analysis of life time data for repairable systems. We started by extending a proportional intensity model which is based on repair history to incorporate the effect of fixed and time varying covariates. We conducted simulation study to obtain the bias and efficiency of the estimates of this model. Following that the performance of the Wald confidence interval estimation technique for the parameters of the models were investigated via a coverage probability study. The models were then applied to a real data set and hypothesis and goodness of fit tests were conducted to see the significance of the model parameters and the suitability of the models. The result of this study demonstrated the ability of the models to capture time trend, repair and covariate effects simultaneously. Next, we extended a log-linear model to accommodate interval failure data assuming an independent inspection process with time varying covariate. Following that we investigated the performance of this model at various sample sizes using tested simulated data. The Wald confidence interval estimation technique was analyzed by the coverage probability study using simulated data. Finally, the result obtained by fitting the real data from conveyor ball bearing failure to several models with interval failure data were compared and analyzed.

PDF FS 2010 51R.pdf Restricted to Repository staff only Download (484kB)

Actions (login required)

View Item