BAYESIAN NETWORKS FOR RELIABILITY ANALYSIS OF TECHNICAL SYSTEM

Abstract

The works to provide reliability assurance and enhancement are most effective at the early stages of life cycle. The common requirement is fast and precise performance of reiterated reliability analysis, as far as the system design development and the complete data accumulation on system components reliability. Recently, Bayesian networks has been introduced in reliability engineering. Bayesian networks are a promising tool for reliability analysis. Today, Bayesian networks are actively used to fault diagnosis and reliability estimates. In this article, utilization of the Bayesian networks for reliability analysis at the early stages of life cycle is studied. The reliability can be calculated by converting block diagrams or fault trees into Bayesian networks. Fault trees can easily be converted into Bayesian networks. Therefore, in this article, a perspective approach to combine the Bayesian networks mechanism and Fault tree analysis method is studied for reliability assessment of a technical system. The system is an electronic fire sensor. As a result, the developed model of Bayesian networks provides the estimation of failure probability at the conditions of real-life scenarios of system behavior; the calculation of the probability of no-failure operation of the system under control and its sub-systems; the detection of the elements that contribute the most to the failure probability. The Bayesian networks model can be easily built from the circuit diagram analysis and the FMEA (Failure Modes and Effects Analysis). In general, this method has the best capability in this real instance compared to other methods.