Induction motor bearing failure diagnosis with ANN and hybrid networks model

This paper proposes five artificial intelligent (AI) methods to determine in- duction motor bearing (IMB) fault diagnosis. In this case, two artificial neural networks (ANN) which are Feedforward Neural Network (FFNN) and Elman Network (EN) and three hybrid networks, which are FFNN with GA (FFGA), EN with GA (ENGA), and adaptive network-based fuzzy inference system (ANFIS) are examined to classify IMB failure. The vibration signal used in this work is obtained from Case Western Reserve University website. During preprocessing stage, the vibration signal have been converted from time domain into frequency domain through Fast Fourier Transform (FFT). Then, an enveloping method is used to eliminate the high frequency components from the vi- bration signal. Subsequently, a set of 16 features from vibration and preprocessed signal is extracted. In order to reduce the size of data, a distance evaluation technique is used as features selection. Lastly, during fault diagnosis all AI methods are examined whose results are compared and conclusions are drawn.