Cooling Fan Combined Fault Vibration Analysis Using Convolutional Neural Network Classifier

In this paper, an application of Convolutional Neural Network (CNN) to detect a predefined fault in vibration signal without any feature extraction. The vibration signal, after being normalized, is converted into a 2-D data called vibration image, and these images are passed in the CNN as input to detect whether there is a fault or not. Experiments are carried out with bearing data from the cooling Fan of a cement oven in CILAS-Biskra. Tests are done using different image sizes, and different training/testing data sets.