Bayesian and degradation data based zero-failure accelerated life demonstration test for electrical products in high speed train system

The electrical components are widely used in high speed trains. It is very important to verify their reliability with a lower cost. This paper proposed a Bayesian zero-failure accelerated life demonstration test with optimum design method to reduce the sample size under the cost constraints. This test design method takes consideration of the sampling risk, extrapolation risk and the acceleration risk. The acceleration risk is introduced to assess the risk caused by the uncertainty of the acceleration model. A Bayesian model is proposed to utilize the historical experiment statistic in order to reduce the sampling risky by means of the Gibbs algorithm. Then, a decision rule is carried out under the cost and the risk constraints. An application on the electrical power panel of the high-speed train system is presented. The results verify that the proposed method works efficiently in reducing the sample size under the risk and the cost constraints.