Feature Extraction And Diagnosis Of Typical Fault Signals Of Asynchronous Motor Rotors

For the research of fault diagnosis technology for asynchronous motors,the extraction and identification of fault features will greatly affect the accuracy and effectiveness of fault diagnosis.Today,the use of vibration signals to diagnose motor faults is a common method.According to the different characteristics of the motor vibration signal under different conditions,the motor vibration signal is diagnosed.Because the traditional signal analysis method is difficult to quantitatively extract the feature vector in non-stationary and non-linear signals,first of all,this paper establishes an acquisition experiment model based on vibration signal in the problem of signal acquisition,and uses the empirical mode decomposition and information of the collected vibration signal.The entropy method is combined to extract feature vectors.Based on the decomposition of multiple eigenmode components,the energy method and the correlation coefficient method are combined to select the principal component of the main mode including the major faults.The same energy method and the correlation coefficient method can be mutually verified and selected.Correctness.After selecting the principal component of the eigenmode,the entropy theory was used to calculate wavelet energy spectrum entropy,singular spectrum entropy,power spectrum entropy,and wavelet packet space feature spectrum entropy in the time and frequency domain.Four types of information entropy are calculated as rotor rotor fault vibration signals as a feature vector group for motor rotor fault diagnosis.Finally,this paper combines the support vector machine(SVM)and evidence theory(DS)to apply motor fault diagnosis,and combines evidence theory to improve the support vector machine.Then,the three typical faults of the rotor of the motor were diagnosed by multi-class support vector machine(SVM).According to the evidence theory,it was soft output in the form of probability,and the signals were fused under different characteristic conditions.This method can improve the accuracy of motor fault diagnosis.To a certain extent,it can predict the fault in advance and achieve real-time monitoring of the fault.