Nonparametric estimation in trend-renewal processes

Abstract

This thesis gives an introduction to stochastic modeling of repairable systems with failure and maintenance data, in particular the nonhomogeneous Poisson process and the trend-renewal process. It is studying kernel-based methods for nonparametric estimation of the trend function of trend-renewal processes and presents a method using weighted kernel estimation. These weights are found by maximization of the likelihood function that they are included in. The method is then tested on both real and simulated data sets.