Research On Vibration Feature Extraction Method Of Generator Faults

Turbine generator is one of the most widely used and important mechanical equipment in electric power production.Vibration is one of the major faults of the turbo generator.Therefore,How to extract the fault features rapidly and accurately from the generator fault vibration signal is always a hot and difficult point in the field of mechanical fault diagnosis.It has important practical significance to study the fast and effective method of generator fault feature extraction,which is of great significance to ensure the safety and economic operation of the unit,and to improve the efficiency of condition monitoring and fault diagnosis.In this thesis,the fault simulation generator MJF-30-6 is used as the research object to analyze the vibration characteristic signal of the stator under normal condition,stator inter-turn short circuit fault.An intensive research about generator fault feature extraction from three aspects: time-frequency analysis;demodulation analysis;de-noising analysis.In the aspect of feature extraction,analysis of the limitations of the use of traditional fourier transform in dealing with non stationary signals.On the basis of this,a time-frequency analysis method is proposed-time wavelet energy spectrum and applied to the field experiment data.The analysis results show that the algorithm can effectively extract the fault characteristics of the generator.In the aspect of demodulation analysis,based on the minimum entropy deconvolution algorithm is one of the main algorithms on the current research,enhanced characteristic vibration signal detection method of generator based on maximum correlated kurtosis deconvolution its application in the fault diagnosis example shows that this method has good demodulation performance,can effectively extract the fault characteristics of vibration signal.In the aspect of de-noising analysis,an improved self-adapted Top-Hat transformation method based on sine-structure element is proposed to enhance the characteristic vibration signal detection of generator.Its application shows that the method is effective not only in reducing the noises but also in enhancing and extracting the features of rotor vibration generator fault.The achievements obtained in this thesis offer a reference for the diagnosis for the rotor short circuit fault and therefore have active meaning.