Gearbox Fault Diagnosis EMD Spectral Kurtosis And SVM

Gearbox has now been widely used in engineering practice, Change in the transmissionof power and speed showed superior performance. But also a prone to frequent failure ofmachinery and equipment, Once a gearbox components on abnormal and do not get timelymaintenance, May not operate properly will cause the entire production line and thus affectthe quality of the final product, So for gearbox fault diagnosis analysis is necessary.Firstly, the basic structure and vibration mechanism ZS-65-type gearbox are described indetail, Which includes an outer ring, an inner ring, cage, the rolling elements and the commonfailure modes and the corresponding tooth of the diagnostic signal analysis method. In thede-noising using wavelet packet decomposition method effectively de-noising the signalprocessing.Secondly, the empirical mode decomposition (EMD) conducted in-depth research, Andon its decomposition occurred during the end effects presents a new method for inhibiting, ARmodel predicts that the use of the data window function extension and combined effects ofsignal-to-end approach to be carried out effectively suppressed, And to improve the accuracyof the subsequent signal processing. In order to be able to find fault information containedIMF component,This introduces a spectral kurtosis parameters, And the effective combinationof EMD and extract the spectral kurtosis up various conditions gearbox characteristicfrequency, Achieved the desired results.Finally, the use of support vector machines for the reconstructed signal after extractingfeature vectors and a normalized signal processing and pattern recognition were classified,and achieved good results, fully verified spectral kurtosis using EMD and combine the two toextract fault eigenvectors effective, good working conditions for all kinds of gearboxclassified.