Fault Detection for Lubricant Bearing with CNN

Bearings are a key part of rotary machines. Damage to bearings has a negative effect on schedule, production operations. Therefore, it is very important that the bearing is diagnosed beforehand. How well a feature is extracted from a vibration signal greatly affects the performance of traditional intelligent fault diagnosis. Traditional intelligent method, however, typically requires extensive domain expertise and prior knowledge. Instead of traditional machine learning algorithms, deep learning algorithms have an ability to learn discriminative representations effectively and accurately from input data. This deep learning model can overcome the disadvantages of traditional intelligent algorithms. In this paper, we will focus on the idea that achieves high accuracy of lubricant oil bearings using conventional vibration signals without additional thermal cameras as possible. For achieving the purpose, we will use CNN (convolutional neural network) which will be used to learn features for fault diagnosis.