A Novel Intelligent Fault Diagnosis Method and Its Application

In order to improve fault diagnosis precision and decrease misinformation diagnosis, rough set theory(RST) and RBF neural network (RBFNN) are introduced to overcome respective deficiency in order to propose a novel intelligence fault diagnosis(NCIRRFD) model and method in this paper. In this NCIRRFD method, the RST as a new mathematical tool is used to process inexact and uncertain knowledge in order to reduce decision tables for obtaining the minimum fault characteristic subset. At the same time, RST is used to serve for pretreatment data so that RBFNN structure is simplified and learning efficiency is improved in order to get an optimized NCIRRFD model for solving the inference complexity. An actual application case is selected to test and verify the proposed NCIRRFD method. The applied results show that the proposed NCIRRFD method can effectively eliminate false and improve the diagnostic accuracy.