IFDS: An Intelligent Fault Diagnosis System With Multisource Unsupervised Domain Adaptation for Different Working Conditions

Intelligent fault diagnosis is an important subject of mechanical system maintenance. Domain adaptation is a method to solve the problem that the model trained on the training set (source domain) is not suitable for the test set (target domain) due to different working conditions in fault diagnosis. In industrial scenarios, there may be multiple-source domains. For this reason, we proposed an intelligent fault diagnosis system (IFDS) with multisource unsupervised domain adaptive network that adapts to single- or multiple-source domains. The proposed method considers the differences between sources and uses source domain data and a small amount of unlabeled target domain data to mine the feature information contained in the data. IFDS uses a feature extractor to learn the feature representations, constructs a domain discriminator for each source domain, and learns domain-invariant features through adversarial training to diagnose target domain faults. The validity of the method in fault diagnosis is verified by various transfer tasks of two public fault-bearing datasets.

I. Introduction

Mechanical equipment has gradually become automated and intelligent, and its internal structure is also more complex now. Once mechanical equipment fails, catastrophic consequences will occur. It is particularly important to diagnose faults in time to ensure the high-precision and high-reliability operation of mechanical equipment. Fault diagnosis aims to detect faults on faulty systems by using the measured values to monitor and analyze the status of the machine, which requires prior knowledge of skilled experts [1]. The expansion of mechanical system operating data promotes the development of data-driven fault diagnosis methods. Signal processing and artificial intelligence are two main methods of data-driven fault diagnosis [2]. Features are extracted from vibration signals through wavelet transform and other signal processing techniques. Machine learning methods, such as restricted Boltzmann machine [3], extreme learning machine [4], and Bayesian network [5], are used for diagnosis. In the past few years, deep learning methods have also shown excellent learning capabilities in many aspects [6], including fault diagnosis. The advantage of deep learning is that it can automatically extract the representation features of the raw data [7]. Xie et al. [8] studied deep neural networks to realize fault diagnosis of motor bearings. Yu et al. [9] proposed a fast deep graph convolutional network for gearbox fault diagnosis. Using raw data from multiple sensors, Shao et al. [10] proposed a motor fault diagnosis method based on a convolution neural network (CNN).