A life prediction approach based on integrated failure information

In this paper a Bayesian method is introduced to evaluate products´ life by integrating the failure information in field and prior failure information collected from various sources. Calibrators are used to calibrate the difference between failure information in field and prior failure information, then Bayesian approach can be used to integrate failure data. The posterior distributions of unknown parameters can be obtained through statistical inference procedure, which is carried out through Markov chain and Monte Carlo (MCMC) method. Normal distribution and 2-parameters Weibull distribution are discussed based on the fusion model established, and simulation examples are performed to illustrate the use of proposed method. The orthogonal analysis shows the influence of different chosen values to prior means of unknown parameters on prediction results.