A Comparative Analysis of Accelerated Life Testing Models

• Main Authors: Shiau, Ming Nan, 蕭名男
• Other Authors: Pan J. N.
• Format: Others
• Language: zh-TW
• Published: 1998
• Online Access: http://ndltd.ncl.edu.tw/handle/12673931005883475005

碩士 === 國立成功大學 === 統計學系 === 86 === The purpose of the reliability life test is to ensure that the product will operate without failure during its expected life span. Usually, the time to failure distribution of random samples is assumed to follow a certain distribution. However, with today''s highly reliable components, we are often unable to obtain a reasonable amount of test data under the normal usage. Instead, most high-tech companies are adopting the Accelerated Life Testing Models (ALTM) by forcing components to fail under much higher environmental stresses; such as: higher temperature, higher pressure levels. By testing this way, we get failure data that can fit into various life distribution models, with relatively small sample sizes and feasible test times. Generally speaking, there are two types of life testing situations. When a truncation time is specified and the test items are replaced as they fail, this is called Type I censoring. When the first r out of n failures occur and the failured items are not replaced, this is called Type II censoring. With regards to the above testing data, this research will first investigate three types of Accelerated Life Testing Models as listed below ： (1) Statistics-based Parametric Models: Four common distributions will be utilized to establish the ALTM if the failure time of the component is difficult to determine based on physics or chemistry principles. (2) Statistics-based Non-parametric Models: the Kaplan-Meier method will be used when the distribution shapes of failure-time data are largely unknown. (3) Physics/Experimental-based Models which utilize the effect of the applied stresses on the failure rate of the units under testing. Such as Arrhenius Models, Eyring Models, Inverse Rule Power Models, etc. Systematic comparison have been made among the above three types of ALTM. Moreover, a decision flow chart has been developed to show the proper selection of ALTM under different conditions and the accurate estimation of the life parameters/MTBF for the components when used under normal stresses. Finally, lower limits of a 95 percent confidence are used as a criteria to conservatively select to the appropriate parameters for the tested components. The above procedure is simplified by the Matlab computer program. Numeric examples have also been given to demonstrate the application of the above program.