一种改进的DSmT在电机振动故障诊断中的应用

Multi-vibration transducers were installed on sensitive sections of motor. Data acquisition system was formed. Multi-source information fusion method was used to diagnose the motor with early vibration fault symptoms. Method based on the intrinsic mode functions (IMF) energy entropy was used to extract signal's feature. Generalized basic belief assignment function was constructed on basis of RBFN outputs, Multi-source information fusion method based on improved DSmT was used to fuse different evidences and make final decision. DST and DSmT have something illogical in dealing with high conflicting evidences. A weight assignment method was adopted considering number, confidence and support level of different evidences. Proportion conflict reallocation rule 4 based on DSmT was used to fuse. Results indicate this method advanced the reliability, solve the conflict among evidences much reasonably. It has theoretical and practical value.