Condition Monitoring of Electric Motor with Convolutional Neural Network

Abstract

Safe, efficient and uninterrupted operation of machine requires continuous monitoring of its health and modern autonomous smart factory demands a Condition Monitoring (CM) process without direct human involvement. Deep Learning (DL) algorithms have shown great success of learning directly from data in various real life applications and recently it become also popular in CM researches but still detail clarification of selecting the DL design and its relevance to learn the features from data are often missing. This paper shows a DL algorithm - Convolutional Neural Network (CNN) to CM of an Electric motor from its external vibration. The output of the deep layers of the learned model is analyzed to explain how the model extract features of raw vibration input and do the classification of different conditions.