A reliability approach to estimating dynamic performance of system using fuzzy theory

摘要

This paper presents a logical approach to estimate system state and its reliability according to the status set of measured sensing signals. The estimated approach is developed based on the combination of Bayesians theory and fuzzy set theory. First, system state is classified into four stages, namely normal operation, slight degradation, function bias and failure. The occurred probabilities for these states are evaluated based on the fuzzy information of system reliability and maintainability of similar equipments. Second, several sensing signals are considered and the signal range of each is also divided into four levels for indicating the system states from normal to failure. An estimated process based on fuzzy set theory is developed to evaluate the possibly occurrence of the signal levels on each system state when system is running. Integrating the estimated results, the conditional probabilities of system states with respect to signal levels can then be calculated by Bayesian rule. The results are further used to compute system reliability by incorporating the successful probabilities of system surviving which are also evaluated based on fuzzy set theory. Based on the proposed approach, the performance state and the dynamic reliability of a system can be validly assessed when the status set of the sensing signals is acquired. The results would support the development of system monitoring in time and provide an early warn for the system degradation.