Redundancy System Fault Sampling Under Imperfect Maintenance

When conducting simulation for evaluating complex system reliability or availability, method of random sampling is applied to simulate fault occasion of complex system. Current research of fault sampling generally assumes maintenance activity restores systems to “good-as-new” and “bad-as-old” without considering “partly good” situation. However, system usually keeps its function through imperfect maintenance that restores system to “partly good”: the system after maintenance will not as good as new, but younger. A system fault random sampling method under condition of imperfect maintenance must be presented in order to enhance simulation model credibility and correctness. When reliability structure of system is in the form of redundancy, the fault sampling method under condition of imperfect maintenance is difficult because not only redundancy system fault time isn’t exponential distribution but also that the state transition process of redundancy system is hard to model and steady state probability of each component in the redundancy system is hard to determine .So in this paper , a generally repairable parallel system with non-identical components: which is a common form of redundancy is considered, then a fault random sampling method for this kind of redundancy system under condition of imperfect maintenance according to monte carlo simulation principle is presented. The characteristics of this method mentioned above is that Markov chain embedded within this method is employed to model the state transition process of redundancy system and to determine steady state probability of each component in the system. Fault occasions of redundancy system under imperfect maintenance can be simulated and fault components in the system can be determined via the fault sampling method. These are novel contributions made in this paper. Finally, a numerical case using fault sampling method for a redundancy system under imperfect maintenance is given and the validity and feasibility of fault sampling method is verified.