Bayesian approach to determine the test plan of reliability qualification test

In view of the shortage of large sample size in Reliability Qualification Test, a Bayesian model is introduced in this paper. Taking advantage of the historical information and expert experience as the prior information, the reliability of the product at current stage is estimated by making use of the new Dirichlet distribution and filed test information in reliability growth testing. Because of the parameters of the new Dirichlet distribution having no physical meaning, to obtain the parameter estimators of prior distribution, we adopt the ML-II method to transfer the failure information at different stages. Considering the balance of consumer’s risk and producer’s risk, the minimum testing plan is scheduled. This method could be regarded as a supplementary program in reliability testing for qualification. An example illustrates the contrast testing test sample size between the proposed method and the standard scheme which validate the proposed approach.