Bearing Fault Diagnosis of Induction Machine Based Empirical ModeDecomposition Energy and Entropy

In this paper, according to the non-stationary characteristics of rotatingmachinery vibration signals, a method to the detection and classification of rollingelementbearing faults of induction machine using Empirical Mode Decomposition (EMD)Energy and Entropy is proposed. Firstly, the vibration signals are decomposed into afinite number of stationary intrinsic mode functions (IMFs), then the EMD energy andentropy are calculated. The analysis results from EMD energy and entropy of differentvibration signals show that these parameters will change in different frequency bandswhen bearing fault occurs. Therefore, to identify roller bearing fault class, the bestfeatures extracted from a number of IMFs that contained the most important faultinformation are selected using a wrapper algorithm that use the Adaptive Neuro-FuzzyInference System ANFIS classifier to evaluate subsets of features. The final features couldserve as input vectors of trained ANFIS. The analysis results from roller bearing signalswith inner-race , out-race and ball faults show that the proposed diagnosis approachbased on ANFIS by using EMD to extract the energy and entropy values of differentstationary intrinsic mode functions as features can identify roller bearing fault patternsaccurately and effectively.