A novel feature extraction method using deep neural network for rolling bearing fault diagnosis

• Published in: The 27th Chinese Control and Decision Conference (2015 CCDC) pp. 2427 - 2431
• Main Authors: Weining Lu, Xueqian Wang, Chunchun Yang, Tao Zhang
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.05.2015
• Subjects: Artificial neural networks; Data mining; Deep Neural Network; Fault diagnosis; Feature extraction; Rolling bearings; Training
• DOI: 10.1109/CCDC.2015.7162328
• ISSN: 1948-9439

Rolling bearing fault diagnosis has received much attention because of its importance for the rotatory machinery. Feature extraction is the crucial part of rolling bearing fault diagnosis, which determines the diagnosis performance greatly. However, features extracted by many available methods cannot guarantee the sensitiveness to every interested fault category, which leads to incomplete diagnosis results and ability absence of handling with the situation that unknown-category fault appears. To solve this issue, the feature extraction method based on deep neural network (DNN) is proposed to extract a meaningful representation for bearing signal in this article. DNN is a new kind of machine learning tool with strong power of representation, which has been utilized as the feature extractors in lots of practical applications successfully. Afterwards, the effectiveness of this proposed approach is presented by using the actual rolling bearing data.