Bearing Fault Detection based on Few-Shot Learning in Siamese Network

Abstract

This paper executes bearing fault diagnosis with little data through few-shot learning. Recently, deep learning-based fault diagnosis methods have achieved promising results. In previous studies, fault diagnosis requires numerous training samples. However, in manufacturing, it is not possible to obtain sufficient training samples for all failure types under all working conditions. In this work, we propose a Few shot learning-based rolling bearing fault diagnosis that can effectively learn with limited data. Our model is based on the siamese network, which learns to use the same or different class of sample pairs.