Wavelet Packet - IC Analysis Method For Gearbox Fault Diagnosis

Gearbox is one of the most common transmission devices in rotating equipments. Whether the operation is normal or not in the mechanical equipments has great influence on operating efficiency, operating effect and the working life. Once a failure occurs in the gearbox, some unpredictable consequences would be caused. Vibration signals in the gearbox contain abundant gearbox fault information, but since the early fault signals in the gearbox are very weak, which are easy to be flooded by vibration signals of other components and noise outside. Therefore, extracting fault information from the vibration signals of the gearbox accurately is of great significance on the gearbox fault diagnosis.The fault of gear in the gearbox is studied in this paper. The wavelet package and ICA analysis method is used to extract gear fault information from the vibration signals in the gearbox. Main work is as followed:(1) An improved algorithm of ICA in super Gaussian signals is used based on the characteristics of vibration signals in the gearbox. This method can be used in diagnosis and separation of the vibration signals which are super Gaussian signals. The result of the simulated analysis suggests that the effect of the method is remarkable.(2) A method of wavelet packet-ICA analysis in this paper is proposed in fault diagnosis in gearbox. Firstly, the original vibration signals are de-noised by the wavelet packet analysis. Secondly, the vibration signals contains the running state of the gear are separated by the improved ICA method introduced in this paper. Thirdly, the wavelet packet envelope spectrum analysis is applied in the vibration signals separated, when the fault is detected, energy eigenvalues of the signals in different broadband can be obtained by the wavelet packet frequency band decomposition technique. Finally, the energy eigenvalues are poured into SVM which can be used to classify the gear faults.(3) The original vibration signals are collected by the experiment platform of gearbox fault diagnosis. With the MATLAB platform, the fault diagnosis and separation algorithm based on the wavelet packet-ICA analysis can be realized. After compare and analysis, the method applied in this paper is verified of validity, which provides an effective way to diagnosis in the gearbox fault diagnosis.