Evaluation methods of storage reliability for relay based subsystems under the conditions of small samples

Component performance degradation due to the effects of storage environmental conditions causes the degradation of system's performance. Thus, a new evaluation method is proposed in this paper to assess the storage reliability of relay based subsystems (RBSs) under the conditions of small samples. And a practical implementation of the method is carried out to manifest the effectiveness. In which, the sensitive components that lead to storage degradation of RBSs' output are determined by simulation based circuit analysis. Degradation data of the testing RBSs as well as the testing relays are observed through storage accelerated degradation testing. With random sampling of these testing relays, 10,000 virtual RBSs are constructed. Based on Simulink simulation and the observed testing data of the sampled relays, degradation data of these virtual RBSs are generated to serve as prior information, with which, the prior estimations of the hyper-parameters are obtained based on expectation maximization (EM) algorithm. A Bayesian theory based formula is studied to compute the posterior estimations of the hyper-parameters with single testing RBS, so as to calculate the posterior expectation of degradation data distribution parameters of RBSs. Finally, the RBSs' reliability curves of both point estimation and lower confidence limit are obtained and shown at the end.