Empirical Bayesian Method Of Accelerated Life Tests

This paper mainly studies the application of empirical bayesian method in the constant stress accelerated life test,and illustrate that it is necessary to use this method to analyze the accelerated life test by comparing the estimation accuracy of bayesian and empirical bayesian methods.Specifically,assuming that the life distribution of the products in each group is Weibull distribution,the failure mechanism of the product is the same,the group effect of each group is a random variable obeying the same distribution,and the acceleration model is the inverse power model,we model and analyze the grouped constant stress accelerated life test.In the case where the test data is timed censored data,we use mixed Gibbs sampling to respectively calculate the bayesian estimation and empirical bayesian estimation of unknown parameters in the model,and uses numerical simulation to verify that the estimation accuracy of empirical bayesian estimation is higher than that of bayesian estimation in the analysis of constant stress accelerated life test.Finally,modeling and analyzing the spacecraft pressure vessel accelerated life test data which Gerstle and Kunz announced in 1983 modeling analysis in the same way,we get the specific speed of the test and the specific life distribution of these pressure vessels,and confirm that empirical bayesian method has higher estimation precision in the analysis of grouped constant stress accelerated life test with stress as acceleration factor.