Education Case on Deep Learning Based Electric Machine Automatic Fault Detection

Automatic fault detection represents a key asset in the predictive maintenance field. Unfortunately, the application of techniques based on Machine Learning (ML) is not widespread in traditional educational curricula. This paper presents a technical lecture to familiarize graduate students and predictive maintenance practitioners with Deep Learning (DL) tools for automatic fault detection. Vibration data from an open source bearing dataset is utilized for training and testing a Convolutional Neural Network (CNN) model. The presented content is developed in a Matlab interface, which is a widespread tool across academic institutions worldwide.