Multi-stage Bayesian reliabilty growth assessment for laser facilities

Abstract

The reliability levels of SG-III component systems are directly relative to the global reliability and stability of SG-III laser facility. The systems consist of mass identical subsystems generally, whose integration process is complicated, and the evaluation of reliability growth is hard to perform directly. Therefore, accounting for subsystems characterized by small samples, hybrid fix modes and inadequate experts experience, a multi-stage Bayesian evaluation method of system reliability growth is proposed. This method integrates AMSAA with the Non-Homogeneous Poisson Process: The failure processes of subsystems in the first stage with enough samples are modeled by AMSAA, and the Non-Homogeneous Poisson Process is assumed to be in accordance with the failure processes at the latter stages where the samples dropped off gradually. And according to the inheritance information of homogeneous subsystems, equivalent conversion methods are used to obtain the growth factors between stages and to solve the problems of lacking experts experience; then the effective transfer of historical failure information and repair information are achieved by combining with the maximum entropy method. Finally, comparative analysis results of numerical case show that this method is suitable for the reliability growth evaluation of the systems of SG-III laser facility, and provides effective ideas of reliability growth evaluation for practical engineering.