Study on Gearbox Gear Fault Diagnosis Based on Wavelet Packet Transform and CEEMDAN and SVD
Accurate fault diagnosis of gearboxes under noise conditions is a difficult problem for gearbox fault di-agnosis.To solve this problem,Firstly,the acquired raw signal is decomposed into wavelet packets,according to the meshing frequency of the faulty gear and select a suitable wavelet packet to reconstruct the signal to obtain a preliminary noise reduction signal.Secondly,CEEMDAN is used to decompose the preliminary noise reduction signal.The correlation coefficient and kurtosis change curves of each IMF component are plotted and the compo-nents with large correlation coefficients are selected for reconstruction.Finally,the signal is further denoised by singular value decomposition,and the final signal spectrogram is compared and analyzed to determine the fault lo-cation and type.The results show that this method can effectively extract the characteristic frequency lines in the specific frequency band according to actual needs,which is better than the results of direct FFT analysis of signals.
fault diagnosisadaptive noise complete set empirical mode decompositionssingular value decom-positionwavelet packet transform