Information Fusion Method for Fault Diagnosis Based on Incomplete Fuzzy Rule Bases

Xu Xiaobin,Ji Yindong,Wen Chenglin
2011-01-01
Abstract:In the fuzzy-reasoning expert system of fault diagnosis,the fuzzy rule base is often incomplete,thus leading to uncertain outputs for some possible inputs.A fusion fault diagnosis algorithm is presented based on the fuzzy rule reasoning and the evidence theory.Firstly,using the random set formalism of evidence and the extension principle of random set,fuzzy inputs are mapped to outputs,then integrated to generate basic probability assignment(BPA) for measuring the uncertainties of outputs caused by the incompletion of system and the fuzziness of inputs.Then,the Dempster combination rule is used to fuse BPA coming from different incomplete rule bases,and faults can be decided by fusion results.Finally,fault diagnosis tests on machine rotor show that the method can effectively improve diagnostic rates.
What problem does this paper attempt to address?