Research On Signal Analysis Methods And System For Fault Sources Of Gearbox

As a transmission device, gearbox is widely used in various mechanical equipments, the diagnosis of the gearbox is very important for the normal running of the equipments. Vibration diagnosis is the key main method for fault diagnosis, but for the complexity of the vibration transmission in the gearbox and difficulty of the end of the local oscillator capture, it is difficult to diagnose the faults of the gearbox. Supported by the Natural Science Foundation Youth Fund project (NO.50205025) and the Natural Science Foundation Fund project (NO.50675194), the methods of independent component analysis based on wavelet packet to capture the fault sources of gearbox was investigated, and the relevant signal analysis system was developed in the environment of LabVIEW combined with MATLAB, Finally the test rig with multi-sources was designed and constructed, the experiment data was analyzed with the former methods. The major contents of the research and the relative chapters of this paper are as follows:In Chapter one, the importance of the fault diagnosis of the gearbox was stated, and the techniques of the fault detection of gearbox were introduced. The vibration signal analysis methods were introduced from time-domain, frequency-domain, modern signal analysis domain, the wavelet analysis and the independent component analysis methods for gearbox fault diagnosis were overviewed. Finally, the main contents, the technology route and the general structure of this paper were given.In chapter two, the mechanism of gearbox vibration generation and transmission was analyzed and the model of gear vibration was carried out. The vibration model of the gearbox faults was studied, the characteristic of the fault vibration signal modulation and aliasing in gearbox was analyzed to help to correct selection of fault separation algorithm, and the Applicability of the independent component analysis based on wavelet packet in the separation of gearbox fault sources was explained. The characteristic of vibration signals of typical gear faults such as cracks, pitting was analyzed which provided a basis for fault feature extraction and fault diagnosis.In chapter three, the theory of the multi-resolution analysis by wavelet, the decomposition and reconstruction of wavelet packet was introduced, and the ranking methods of the Gray code for the wavelet decomposition was discussed, a feature extraction method of gearbox vibration signal based on wavelet packet decomposition and reconstruction was given and simulation signal investigation was also carried out, the miscellaneous frequency phenomenon which was caused by wavelet packet analysis was discussed and analyzed, it was concluded that the generation of the miscellaneous frequency phenomenon was related to different wavelet functions and the sampling frequency.In chapter four, the basic theory and algorithm of ICA were introduced, the high-order ICA algorithm (HOS) and the second-order ICA algorithm (SOS) were given and the advantages and disadvantages of the algorithms was compared by the analyzing of the simulation signals which represented different features of the signal. For instantaneous mixtures and convolutive mixtures models of vibration signals of gearbox, the method of source separation based on the wavelet pre-processing was given for the instantaneous mixtures, and the method of separating sources in the frequency domain based on the wavelet pre-processing was given. And the two methods were proved to be effective for different models through simulation.In chapter five, signal analysis system based on wavelet packet and ICA analysis was developed, the general design idea and overall structure of the system were introduced; the developing tools such as MATLAB, LabVIEW were introduced, and the combination of these tools by COM component technology was discussed, finally the algorithm of the ICA based on COM in MATLAB and the algorithm of wavelet decomposition and reconstruction of wavelet packet based on LabVIEW were obtained.In chapter six, according to the gearbox fault diagnosis, the fault simulation system was established including gearbox multi-sources vibration simulation test platform, vibration control system, data acquisition system and so on. Various experiments such as experiments of motor size , experiments of motor and gearbox were carried out, by analyzing the experimental data, the effect of previous algorithms were proved, and the algorithm based on wavelet packet can be much more effective, which provided effective supports to multi-source analysisIn chapter seven the conclusion and achievement of this research was given, and the further work and the related work suggested.