This study focuses on a special case of accelerated life testing (AL T) called the step-stress experiment and in particular on the simple step-stress model involving only two stresses levels (Refs. 1-2). This article presents a likelihood inference for the multi-sample simple step-stress model under the same set-up, under exponential lifetimes and Type I censoring. The study considers the Iife-testing problem of s independent samples under two stress levels, common in all samples, when the lifetimes are exponentially distributed under both stress levels in all samples whereas the test is terminated at prefixed times that could possibly be different for the s samples. Also, for different time points the stress levels in the s samples are changed. Multi-sample step-stress models were introduced recently by Kateri, Kamps and Balakrishnan (Ref. 3), where such an extension is considered. (16 refs.)
展开▼