The gross performance of a system is decided based on many observed signals. To accurately evaluate system performance from the observed signals, this paper reported an evaluated method of system states and reliability based on a combination of Bayesian method and fuzzy signal information. The states of system were classified into four levels including normal, slight degradation, function deviation and broken down as well as the observed signals each were graded with many levels. Fuzzy numbers is used to construct the correlations between the system states and the signal levels. The event probabilities of the system states over the signal levels are decided using Bayesian evaluation. The related elements in Bayesian evaluation were given by a conversion of fuzzy possibility score for the fuzzy numbers. The prior information of system states were based on the steady-state probabilities of system, which were acquired by Markov transition analysis. The evaluated results are further combined with fuzzy reliabilities of states to rate dynamic reliability of systems. The estimated method provides a quantified assessment for system gross performance from the observed signals which may be helpful in developing on-condition maintenance. Properly, integrating the method into monitoring systems, the system states at any time can be detected in time as a useful information in decision-making for production adjusting or maintenance performing.
關聯:
JOURNAL OF THE CHINESE SOCIETY OF MECHANICAL ENGINEERS
