Fault Diagnosis Based on the Updating Strategy of Interval-Valued Belief Structures

This paper presents the dynamic method for fault diagnosis based on the updating of Interval-valued belief structures (IBSs). The classical Jeffrey's updating rule and the linear updating rule are extended to the framework of IBSs. Both of them are recursively used to generate global diagnosis evidence with the form of Interval basic belief assignment (IBBA) by updating the previous evidence with the incoming evidence. The diagnosis decision can be made by global diagnosis evidence. In the process of evidence updating, the similarity factors of evidence are used to determine switching between the extended Jeffrey's and linear updating rules, and to calculate the linear combination weights. The diagnosis examples of machine rotor show that the proposed method can provide more reliable and accurate results than the diagnosis methods based on Dempster-Shafer evidence theory.