Fault Diagnosis Of The Asynchronous Motor Based On Wavelet Package And LS-SVM

Squirrel-cage asynchronous motors because of its simple structure, reliableoperation, and low price, is widely used in industrial and agricultural production. Butaffected by the poor working environment, frequent starting braking and the influenceof other factors such as force, all kinds of faults and failures often happen on theinduction motor, which will not only damage the motor itself, but also affect theoperation of the whole industrial production. Therefore, diagnosis and analysis ofcommon faults using effective methods has been becoming the current research ofsquirrel-cage asynchronous motor of the mainstream. The traditional diagnosismethods have certain limitations, cannot very well to extract fault features. Articles inthe research object of squirrel cage induction motor stator current signal, in theperspective of characteristic frequency of stator current, is put forward according tothe theory of wavelet packet and support vector machine (SVM) theory as thefoundation of squirrel cage asynchronous motor fault diagnosis methods. The mainwork is as follows:First, on the basis of Analyzing of squirrel cage motor rotor structure, and therotor broken bar fault reason, and fault mechanism, the diagnosis methods are summarized,and the stator current diagnostic methods is proposed; The second, study the wavelettransform and the basic principle of wavelet packet transform, and further put forwardoptimal wavelet noise reduction and frequency band energy method based on thewavelet packet theory, and The validity and reliability of this method is verified bysimulation of MATLAB; Again, Summary analysis of the basic theory of machinelearning, Focus on Support vector machine classification theory and the theory ofleast squares support vector machine is analyzed and the theoretical deduction; Thelast, Diagnosis of stator current signal collected from the fault test platform is theoriginal data, Using the MATLAB programming phase in the above analysis, Gettingthe high accuracy of fault diagnosis.According to the article, the method based on wavelet package and least squaressupport vector machine (SVM) can eliminate noise and obtain the diagnosisinformation very good, accurately classified judgment. And has the very good theoryvalue and application and dissemination value.