Comparative Investigation of Broken Bar Fault Detectability in Induction Motor Through FFT and MUSIC Techniques

Abstract

The purpose of the current study is to detect the squirrel cage induction machine under the Broken Rotor Bars fault (BRBs). In this context, most of the studies were starting by using the Fast Fourier Transform (FFT) technique and applied to the various signals extracted from the motor as the stator phases currents. Unfortunately, FFT technique has some drawbacks such as suffering from the spectral leakage, also, it needs large data points to give clear results about the state of the machine. In addition, most of the induction motors are using speed control devices (the inverters), these devices reduce the effectiveness of FFT because it leads to the appearance of an additional harmonics, and it produces a further spectrum noises. To overcome these limitations, the Multiple Signal Classification (MUSIC) algorithm have been applied to replace the FFT. In particular, MUSIC algorithm allows to minimize the computation of the signal data without losing its diagnostic effectiveness, on the other hand, this technique allows to remove the noise that accompanies the signal. In this paper, the simulation results evidence the robustness of the MUSIC technique to detect the BRBs when the machine is operating under different conditions.