Generating optimal design for ALT experiment with time censoring

Abstract

The computer generated optimal experimental designs have become popular as the computational capabilities of computers grow. For example, the optimal test plans of accelerated life tests (ALT) have received a lot of attentions from academia and practitioners in recent years. In this paper, we discuss a computational method for finding the D-optimal ALT plans through the generalized linear model (GLM) of accelerated life testing problem. A modified version of the point exchange algorithm (PEA) is developed. By comparing this algorithm with the quasi Newton method on an ALT planning problem, we demonstrate the superiority of our method in terms of speed and accuracy.