Induction Motor Bearing Fault Classification Using PCA and ANN

In the present work, the induction motor bearing fault has been classified by using artificial neural network (ANN). For the classification purpose, the features have been extracted from signal in time domain. Two schemes are framed; the first one is scheme 1 in which original recorded signal has been used and calculated statistical parameter. In scheme 2, the signal has been chosen by applying wavelet packet transform (WPT) on original signal to find that node which is rich in information. ANN is then applied to both to classify the faults. In addition, principal component analysis (PCA) has been used for both the schemes to reduce the number of features and consider only significant features. Further, ANN is applied for classification with reduced feature through PCA. All the results are compared together and reveal that ANN classification is better if all features are considered for scheme 2.