Motor Bearing Faults Detection and Classification based on Convolutional Neural Network and Support Vector Machine: A Comparative Study

Abstract

Rotating bearings are one of the widely used components in machinery systems. Bearings are the main reason for the occurrence of faults in rotating machinery systems. Accurate and quick bearings faults detection is important for machinery systems. Nowadays, Deep learning comes up as a very effective artificial intelligence technique. CNN or Convolution neural network is a class of deep neural networks that are used for the diagnosis of faults. Another technique is the support vector machine technique which is a supervised machine learning model which is effectively used for fault classification. In this study, Convolution neural network (CNN) and support vector machine (SVM) algorithm is proposed for fault detection and classification. For the classification of rolling bearing faults, Firstly, vibration signals are converted into time-domain signals and normalization has also been done for achieving better result. A new model is generated for fault classification based on Convolution neural networks and SVM algorithm. To find out the bearing fault and to classify them in real-time a training model can be used. Comparative analysis is done and experimental results show that the CNN model classify with 100% accuracy. To show the effectiveness of the proposed algorithm, the performance is compared with existing literature works. Better results are obtained from the algorithm of CNN than the existing work.