On the motor fault diagnosis based on wavelet transform and ANN

Digital signal processing methods are adopted to carry on the smart diagnosis on the electric motor fault type judgment on MATLAB platform. Firstly, we gathered electrical machinery´s sound signals in different running conditions and de-noised these signals by using wavelet method in time domain and frequency domain. Next, the signal´s energy eigenvector is analyzed and extracted. Finally, the neural network sorter was operated to classify the quantified electric motor fault sound. Several methods are adopted throughout the whole process of de-noising, extraction of the energy eigenvector and neural network recognition. The comparison of these methods is also made so as to select the optimal one for the electric motor fault type diagnosis. The experiments indicated that the smart diagnosis introduced in this article achieved high rate of accuracy in the electric motor fault type recognition based on the noise analysis.