Abstract: For the multistage reliability growth tests with hybrid (instant and delayed) fix modes for complex aeronautic mechatronic systems, an multistage nonhomogeneous poisson process (MS-NHPP) model and its Bayesian analysis are presented. In the model, the multistage test is regarded as a combination of multiple nonhomogeneous poisson process (NHPP) processes with the same starting point but different intensity function parameters that satisfy some ordinal constraints. In the corresponding Bayesian analysis with prior Dirichlet distribution, the posterior expectations, variances and confidence limits of reliability parameters of all test stages are derived by Markov chain Monte Carlo (MCMC) sampling based on the Metropolis-Hastings principle. The example analysis indicates that this method can effectively integrate multistage test data, expertise and historical information with sound applicability to small sample cases, and that it can derive parameter estimates that satisfy ordinal constraints and obtain reliable evaluation results for reliability growth tests.

Key words: reliability theory, reliability growth, multistage test, multistage nonhomogeneous poisson process, Bayesian analysis, Markov chain Monte Carlo