A COMPUTATIONAL FRAMEWORK FOR PRIME IMPLICANTS IDENTIFICATION IN NON-COHERENT DYNAMIC SYSTEMS

Dynamic reliability methods aim at complementing the capability of traditional static approaches (e.g., Event Trees (ETs) and Fault Trees (FTs)) by accounting for the system dynamic behavior and its interactions with the system state transition process. For this, the system dynamics is here described by a time-dependent model that includes the dependencies with the stochastic transition events. In this paper, we present a novel computational framework for dynamic reliability analysis whose objectives are i) accounting for discrete stochastic transition events and ii) identifying the prime implicants (PIs) of the dynamic system. The framework entails adopting a Multiple-Valued Logic (MVL) to consider stochastic transitions at discretized times. Then, PIs are originally identified by a Differential Evolution (DE) algorithm that looks for the optimal MVL solution of a covering problem formulated for MVL accident scenarios. For testing the feasibility of the framework, a dynamic non-coherent system composed by five components that can fail at discretized times has been analyzed, showing the applicability of the framework to practical cases