Abstract
Due to the successful modeling on fatigue lifetime in various fields, the generalized Birnbaum-Saunders (GBS) distribution has been recently explored and extended from different respects. In this article, we study a varying-stress accelerated life test for GBS model to induce an extended GBS (EGBS) distribution on product failure times. We outline the attributes of this highly flexible distribution, present the classical maximum likelihood estimation method, develop a novel goodness of fit technique, and propose a new inference in Bayesian framework. The performance of the methods are assessed through a simulation study under various scenarios of distribution settings and sample sizes. For illustrative purpose, we consider two numerical examples to display the accuracy and efficiency of the proposed Bayesian method over the likelihood-based procedure.