Bearing fault feature information extraction method based on GA-FMD and EK composite index

In view of the rolling bearing vibration signal with large noise is in the lead to the problem of the bearing fault information is difficult to effectively extract of the proposed based on GA-FMD and EK composite index of bearing fault characteristic information extraction method. Firstly, in order to overcome the shortcoming of feature mode decomposition (FMD) needing to rely on human experience to set the key parameters, a genetic optimization algorithm (GA) based on Teager energy operator as the fitness function was proposed to adaptively determine the input parameters of FMD. Then, according to the composite screening index EK(EHNR-K) composed of harmonic ratio and kurtosis, several modes of FMD with parameter optimization are decomposed, and the mode with the largest EK value is selected as the intrinsic modes with sensitive fault characteristic information, so as to remove the noise-dominated mode and redundant mode. Finally, the selected intrinsic mode components were reconstructed and the envelope spectrum was analyzed to realize the feature extraction of signal fault information. The simulation and experiment show that this method can extract bearing fault effectively and distinguish bearing fault effectively compared with other traditional decomposition methods.