Optimal Design of Multiple Stresses Accelerated Life Test Plan Based on Transforming the Multiple Stresses to Single Stress

• Published in: Chinese journal of mechanical engineering 2014-11, Vol.27 (6), p.1125-1132
• Main Authors: Gao, Liang, Chen, Wenhua, Qian, Ping, Pan, Jun, He, Qingchuan
• Format: Article
• Language: English
• Subjects: Accelerated life tests; Accelerated tests; Deviation; Electrical Machines and Networks; Electronics and Microelectronics; Engineering; Engineering Thermodynamics; Estimating; Estimation; Heat and Mass Transfer; Instrumentation; Machines; Manufacturing; Mathematical models; Mechanical Engineering; Optimization; Parameter estimation; Planning; Power Electronics; Processes; Quantiles; Straight lines; Stress concentration; Stresses; Theoretical and Applied Mechanics; 优化设计; 加速寿命试验; 压力水平; 多重; 应力水平; 模型参数; 测试计划; 试验方案
• ISSN: 1000-9345; 2192-8258
• DOI: 10.3901/CJME.2014.0826.141

For planning optimum multiple stresses accelerated life test plans, a commonly followed guiding principle is that all parameters of the life-stress relationship should be estimated, and the number of the stress level combinations must be no less than the number of parameters of the life-stress relationship. However, the general objective of an accelerated life test（ALT） is to assess thep-th quantile of the product life distribution under normal stress. For this objective,estimating all model parameters is not necessary, and this will increase the cost of test. Based on the theoretical conclusion that the stress level combinations of the optimum multiple stresses ALT plan locate on a straight line through the origin of coordinate, it is proposed that a design idea of planning the optimum multiple stresses ALT plan through transforming the problem of designing an optimum multiple stresses ALT plan to designing an optimum single stress ALT plan. Moreover, a method of planning the optimum multiple stresses ALT plan which can avoid estimating all model parameters is established. An example shows that, the proposed plan which only has two stress level combinations could achieve an accuracy no less than the traditional plan, and save the test time and cost on one stress level combination at least; when the actual product life is less than the design value, even the deviation of the model initial parameters value is up to 20%, the variance of the estimation of thep-th quantile of the proposed plan is still smaller than the traditional plans approximately 25%. A design method is provided for planning the optimum multiple stresses ALT which uses the statistical optimum degenerate test plan as the optimum multiple stresses accelerated life test plan.