Decision network model for vibration fault diagnosis of steam turbine-generator set based on rough set theory

Abstract

Redundancy and inconsistency are universal features of the turbine vibration fault diagnosis. If we can provide a solution to the problem, it should be very meaningful that the fault diagnosis data included redundant and inconsistent information could be used to decision-making rules of fault diagnosis. In this paper, the model was achieved through constructing a network of fault diagnosis decision-making, which had the different levels. According to the nodes of network with various levels, we could get the diagnostic decision-making rules with the tidy length and compact number. On the basis of a given confidence level, the concept of rule coverage was introduced. So the noises were effectively filtered out and the extraction efficiency of diagnosis rules was improved. In the event that the fault diagnosis was incomplete, the relatively satisfied diagnosis conclusions could also be given.