The Research Of Fault Diagnosis Method For Motor Bearings Based On Mulitaper Higher-Order Spectrum

In industrial enterprise induction motor is the most widely used in drag, the failure of t induction motor will directly affect the normal operation of the line, however, the bearing fault occurred in the highest probability. This paper mainly studies the induction motor bearing fault diagnosis method.At present,based on current characteristic analysis diagnosis fault get more extensive attention, This method established function between motor stator current and the frequency characteristic of the common faults. Based on the characteristics of frequency can detect motor bearings several common faults.When the motor run in different state,stator current may contains different fault characteristic frequency,but the amplitude of Fault characteristic frequency of the signal is very small,so the amplitude of the fault characteristic frequency in power spectrum is very small.Because the main components of the stator current is fundamental harmonic.however,it contains third,forth,fiveth harmonic and background noise coming from grid, we should improve the resolution ratio of the analysis to distinguish fault feature frequency,if we do this,there are several fault peaks so that analysis, results drop.According to frequency characteristics of the stator current,which carries the fault information,there are two methods:Wavelet decomposition and spectral analysis.But the two methods only used frequency information,ignored phase information.Bispectrum estimation not only use the stator current signal frequency information but also use the phase information to ensure motor fault feature by analyzing the coupling relation between fault feature frequency and fundamental harmonic.So bispectrum estimation is suitable for analyzing the nongauss signal submerged in gauss noise.How to extract the characteristics of weak periodic signals under the background of color noise? This paper use the higher order accumulation,which own the ability to restrain colored gauss noise,and use the ability of mulitaper method which can extract weak signal feature, to propose new feature extraction method based on higher order mulitaper estimation.Using mulitaper bispectrum estimation analysis the phenomenon of nonlinear phase coupling between fault feature frequency and fundamental harmonic, Realize the recognition of the weak fault feature under the strong signal characteristics. Through the experimentation,this method can effect to realize the recognition of motor bearing fault feature.