A new data-driven intelligent fault diagnosis by using convolutional neural network

Along with the boosting of big data in manufacturing, applying the data-driven analysis method to support the intelligent fault diagnosis becomes a new trend. Recently, deep learning emerged as a potential artificial intelligence technique. It can obtain the features of raw data automatically, which provides a new way to reduce the expert's bias as more as possible and to mine the inherent relationships hidden in data. Convolutional neural network (CNN) is a promoting kind of deep learning. A new intelligent fault diagnosis method based on CNN is proposed in this paper. Firstly, a transformation from signals to images is investigated to deal with the raw signal data in a simple way. Then these images are trained by CNN. The proposed method is tested on the motor bearing dataset from Case Western Reserve University. The results show that the proposed method achieves as high as 99.51% of the prediction accuracy. The good performance of proposed method is also proved by comparing it with other deep learning methods and traditional methods.