Linear independent increment process with linear standard deviation function for degradation analysis

Abstract

Degradation test is a feasible means to assess the failure time distributions of complex systems and highly reliable products. Generally speaking, the degradation process is essentially a continuous state random process. Motivated by the independent increment process theory, we propose a degradation analysis model, in which degradation is represented by an independent increment process with linear mean and standard deviation functions (quadratic variance function). A one-stage method is further developed to estimate the model parameters and failure time distribution. The methodology has been implemented in the analysis of GaAs laser degradation. The comparative results illustrate that the proposed method can be considered good and show a promise for future applications.