Neural Network Classification and Diagnosis of Broken Rotor Bar Faults by Means of Short Time Fourier Transform.

In this paper an experimental study of classification and diagnosis of different number of broken rotor bars and broken end-ring in the three-phase squirrel cage induction motors is presented. Six different faulted rotors are investigated. These faults are one, two, three broken bars, broken end-ring, a bar with high resistance and healthy rotor. The base structure of the study consist of current signal analysis (CSA), feature extraction, Artificial Neural Network (ANN) and diagnosis algorithm. The motor current signal is used for obtaining of effects of broken bars and end-ring in the rotor. To get sight of the effects the current signal that is in the time domain is transformed time-frequency domain via Short Time Fourier Transform (STFT). And the spectrums are averaged and normalized on the time axis. The rotor cage faults are classified with ANN by using these spectrums. And result matrixes of ANN are considered improved decision structure. Thus the faulted rotors are diagnosed at 100% accuracy and classified 98,33% accuracy. [ABSTRACT FROM AUTHOR]