A fault diagnosis method of bearing using energy spread spectrum and genetic algorithm

Considering the shortcomings of the traditional energy spectrum algorithm applied to the rolling bearing fault diagnosis, which can only represent the tendency of fault feature transformation with a certain scale, but not adjacent scales contained. In this paper, we propose a fault diagnosis method of rolling bearing based on Support Vector Machine, combining energy spread spectrum and genetic optimization The extracted signal is denoised and decomposed using wavelet packets, the energy spectrums and energy spread spectrums are calculated based on the decomposed different frequency signal components. The genetic algorithm is used to select the important parameters of the Support Vector Machine and bring the determined parameter values into the Support Vector Machine to generate the GA-SVM model. Then, energy spectrums and energy spread spectrums are inputted into GA-SVM as the characteristic parameters for identification. The experimental results show the two new points of energy spread spectrums and GA-SVM improve the diagnostic rate by up to 28.5 %, it can effectively improve the fault recognition rate of the rolling bearing.