The Construction of Degradation Trend using the "Random-Effect" Models

In this paper, we compare different ways to construction of the degradation models with random effects. Such models are based on the degradation index's dependence on time, which is called trend in this work. Using the constructed model, we can predict time of the failure, in other words, non-failure time of the tested items. The two ways are considered: it is a Wiener degradation model with covariates, which includes the normal distribution as a distribution of the increments of the degradation index, and semi-parametric model with random effects. Using the computer simulation methods, we conduct the research on accuracy of the trend construction, which coincides the considering models, regarding true trend function.