Reliability modeling of two-phase inverse Gaussian degradation process

Abstract

This paper discusses the reliability evaluation of the two-phase model with the inverse Gaussian (IG) process. In the two phases, the degradation paths are supposed to follow the IG process with different parameters. To represent the subject-to-subject heterogeneity, the change points and the model parameters of different devices are set to be different. For each device, the change point is detected based on the Schwarz information criterion (SIC), and the unknown parameters are obtained by utilizing the maximum likelihood estimation (MLE) approach. Furthermore, the reliability function of each device under the discussed two-phase IG model is also computed. Finally, an example of liquid coupling devices (LCDs) is presented to validate the proposed model, and it can be found that the proposed model fits this data set well.