HOSA For Vibration Signals Of Rotating Machinery

This paper detailedly expounds theories of Higher Order Statisties Analsys(HOSA). There are a series of simulation experiments designed on this basis to verify the theories of HOSA.By using bispectrum,bicoherence spectrum and 1(1/2)-dimension spectrum, we analyze abnormal vibration signals of turbine,vibration signals of the gear pitting failure and vibration signals of the gear crack failure.We can get a conclusion that using HOSA will bring stable performance,high sensitivity,and effectively inhibit Gaussian noise of vibration signals.Meanwhile we can also analyze the phase-change process of the failure in its development.Then there will be more effective fault feature extraction tools than classic methods.And it will be propitious to the latter fault diagnosis.