Two-Stage Maximum Likelihood Estimation Procedure for Parallel Constant-Stress Accelerated Degradation Tests

The parallel constant-stress accelerated degradation test (PCSADT) is a popular method used to assess the reliability of highly reliable products in a timely manner. Although the maximum likelihood (ML) method is commonly utilized to estimate the PCSADT parameters, the explicit forms of the ML estim...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on reliability 2021-06, Vol.70 (2), p.446-458
Main Authors: Wu, Cheng-Hsun, Tsai, Tzong-Ru, Lee, Ming-Yung
Format: Article
Language:English
Subjects:
Accelerated relationship law
Accelerated tests
Acceleration
Asymptotic methods
Degradation
Fisher information
Life estimation
Light emitting diodes
Maximum likelihood estimation
Normality
parallel constant-stress accelerated degradation test
Parameter estimation
Reliability
Reliability analysis
Reliability engineering
Stress
time-transformed model
two-stage estimation
Wiener process
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The parallel constant-stress accelerated degradation test (PCSADT) is a popular method used to assess the reliability of highly reliable products in a timely manner. Although the maximum likelihood (ML) method is commonly utilized to estimate the PCSADT parameters, the explicit forms of the ML estimators, and their corresponding Fisher information matrix are usually difficult to obtain. In this article, we propose a two-stage ML (TSML) estimation procedure for a time-transformed model. In the proposed procedure, all the TSML estimators not only have explicit expressions but also possess consistency and asymptotic normality. Hence, this method is tractable for reliability engineers. Furthermore, the TSML estimators can provide constructive information about the unknown accelerated relationship law. The proposed method is also applied to analyze light-emitting diode data and compare the performance of our estimation procedures with the ML method via simulations.
ISSN:0018-9529
1558-1721
DOI:10.1109/TR.2021.3053312