A NEW WAY TO ESTIMATE THE PARAMETERS IN THE PROGRESSIVE STRESS ACCELERATED LIFE TESTING

Among the three types of accelerated life testing, that is, the constant stress accelerated life testing, the step stress accelerated life testing, and the progressive stress accelerated life testing, the last one is the most effective and economical way to get the failure times of products in a short period of time. In this artiele a deficiency inthe classical approach is pointed out and a new method for the statisticul analysis of lifedata, especially for electronic products, under progressive stress (voltage) V(t)=Kt is given. Using the Bayesian method, for certain choices of the prior distribution, several forms of estimators of the parameters in the noraml stress We|bull distribution and the htverse power law model are derived. In particular, when the new shape parameter computers. The formulations are given for the general case and are illustrated for a special sitnpio case. comparison with the classical approach using pseudorandom dataas well as real data on solid tantatum electrolytic capacitors shows that the proposed method is better and more effective.

A new way to estimate the parameters in the progressive stress accelerated life testing

Among the three types of accelerated life testing, that is, the constant stress accelerated life testing, the step stress accelerated life testing, and the progressive stress accelerated life testing, the last one is the most effective and economical way to get the failure times of products in a short period of time. In this article a deficiency in the classical approach is pointed out and a new method for the statistical analysis of life data, especially for electronic products, under progressive stress (voltage)V(t) =Kt is given. Using the Bayesian method, for certain choices of the prior distribution, several forms of estimators of the parameters in the normal stress Weibull distribution and the inverse power law model are derived. In particular, when the new shape parameter of the distribution under progressive stress is given (or estimated in advance), the closed forms of these estimators are relatively simple and are numerically accessible on computers. The formulations are given for the general case and are illustrated for a special simple case. Comparison with the classical approach using pseudorandom data as well as real data on solid tantatum electrolytic capacitors shows that the proposed method is better and more effective. 1991MR Subject Classification: 62N05.62F. research was supported by the National Natural Science Foundation of China and the Science and Technology Development Foundation of Shanghai Higher Learning.