Research advances in fault diagnosis and prognostic based on deep learning

Abstract

Aiming to condition based maintenance for complex equipment, numerous intelligent fault diagnosis and prognostic methods based on machine learning have been researched. Compared with the traditional shallow models, which have problems of lacking expression capacity and existing the curse of dimensionality, using deep learning theory can effectively mine characteristics and accurately recognize the health condition. In consequence, fault diagnosis and prognostic based on deep learning have turned into an innovative and promising research field. This paper gives a review of fault diagnosis and prognostic based on deep learning. First of all, a brief introduction to deep learning architecture and the framework of fault diagnosis based on deep learning is described. Second, tracking describes the latest progress of fault diagnosis and prognostic based on deep learning in chronological order. In this section, the deep learning methods used in fault diagnosis and prognostic are discussed, including Deep Neural Network (DNN), Deep Belief Network (DBN) and Convolutional Neural Network (CNN). Then the engineering application fields are summarized, such as mechanical equipment diagnosis, electrical equipment diagnosis, etc. Finally, this paper indicates the potential future research issues in this field.