Application of deep neural networks in a mobile application for classifying failures of an induction motor

The article presents the use of a deep neural network to classify failures of an induction motor. Failures are related to inter-turn short-circuits occurring in the stator circuit. The classification is applied as a mobile application using the Intel Movidius Neural Compute Stick. The state assessment was made on the basis of a database containing the results of continuous wavelet analysis of the torque waveforms of the motor for a different number of shorted turns. Various database configurations for the neural network used in the application have been considered.