DocumentCode
3395869
Title
Optimum plans for step-stress accelerated life tests at higher usage rates
Author
Yang, Guangbin
Author_Institution
Ford Motor Co., Dearborn, MI
fYear
2008
fDate
28-31 Jan. 2008
Firstpage
255
Lastpage
260
Abstract
The life of some products is measured by usage, e.g., mileage and cycles. In accelerated testing, these products are usually run at elevated stress levels and higher usage rates to shorten test length. The increase in usage rate may prolong or shorten usage to failure. In this paper, a model that relates life to stress and usage rate is described. For the model, this paper develops 3-step step-stress optimum test plans, which choose the optimal step-change times by minimizing the asymptotic variance of the estimate of the log mean life at a design stress and usual usage rate. The sensitivity of the test plans to preestimates is evaluated, and the results show that the plans are robust. This paper also presents the maximum likelihood estimates of model parameters and the mean life at the design stress and usual usage rate, and their confidence intervals. The proposed method is applied to estimate the reliability of a compact solenoid valve.
Keywords
life testing; maximum likelihood estimation; reliability; sensitivity analysis; solenoids; valves; compact solenoid valve; maximum likelihood estimation; product testing; reliability; sensitivity analysis; step-stress accelerated life tests; step-stress optimum test plan; Data analysis; Life estimation; Life testing; Manufacturing; Maximum likelihood estimation; Parameter estimation; Robustness; Sensitivity analysis; Solenoids; Stress measurement; accelerated life test; optimum plan; sensitivity analysis; step stress; usage rate; usage to failure;
fLanguage
English
Publisher
ieee
Conference_Titel
Reliability and Maintainability Symposium, 2008. RAMS 2008. Annual
Conference_Location
Las Vegas, NV
ISSN
0149-144X
Print_ISBN
978-1-4244-1460-4
Electronic_ISBN
0149-144X
Type
conf
DOI
10.1109/RAMS.2008.4925804
Filename
4925804
Link To Document