Augmented probability simulation for accelerated life test design. (19th May 2017)

• Record Type: Journal Article
• Authors: Polson, Nicholas G.; Soyer, Refik
• Other Names: Soyer Refik guestEditor.; Verdinelli Isabella guestEditor.
• Is Part Of: Applied stochastic models in business and industry. Volume 33:Number 3(2017)
• Page Start: 322
• Page End: 332
• Publication Date: 2017-05-19
• Subjects: Bayesian optimal design -- life testing -- decision theory -- Monte Carlo optimization -- particle methods; Stochastic analysis -- Periodicals; Stochastic processes -- Periodicals; Business mathematics -- Periodicals; Finance -- Mathematical models -- Periodicals; Industrial management -- Mathematical models -- Periodicals; 338.00151923
• DOI: 10.1002/asmb.2256
• Languages: English
• ISSNs: 1524-1904

Abstract

Abstract : Designing accelerated life tests presents a number of conceptual and computational challenges. We propose a Bayesian decision‐theoretic approach for selecting an optimal stress‐testing schedule and develop an augmented probability simulation approach to obtain the optimal design. The notion of a 'dual utility probability density' enables us to invoke the concept of a conjugate utility function. For accelerated life tests, this allows us to construct an augmented probability simulation that simultaneously optimizes and calculates the expected utility. In doing so, we circumvent many of the computational difficulties associated with evaluating pre‐posterior expected utilities. To illustrate our methodology, we consider a single‐stage accelerated life test design; our approach naturally extends to multiple‐stage designs. Finally, we conclude with suggestions for further research. Copyright © 2017 John Wiley & Sons, Ltd.