Motor fault diagnosis based on deep learning frameworks has gained much attention from academic research and industry to guarantee motor reliability. Those methods are commonly under two default assumptions: 1) massive labeled training samples and 2) the training and test data share a similar distribution under unvarying working conditions. Unfortunately, these assumptions are nearly invalid in a real-world scenario, where the signals are unlabeled and the working condition changes constantly, resulting in the diagnosis models of the previous studies that always fail in classifying the unlabeled data in real applications. To deal with those issues, in this paper, we propose a novel feature adaptive motor fault diagnosis using deep transfer learning to improve the performance by transferring the knowledge learned from labeled data under invariant working conditions to the unlabeled data under constantly changing working conditions. A convolutional neural network (CNN) is adopted as the base framework to extract multi-level features from raw vibration signals. Then, the regularization term of maximum mean discrepancy (MMD) is incorporated in the training process to impose constraints on the CNN parameters to reduce the distribution mismatch between the features in the source and target domains. To verify the effectiveness of our proposal, data from the motor tests of European driving cycle (NEDC) for simulating the real working scenario and the motor tests under invariant working conditions are, respectively, conducted as the target domain and the source domain. The results show that the proposal presents higher diagnosis accuracy for the unlabeled target data than other methods, and it is of applicability to bridge the discrepancy between different domains. INDEX TERMS Motor fault diagnosis, transfer learning, domain adaptation, convolutional neural network (CNN).
Data-driven machinery fault diagnosis has gained much attention from academic research and industry to guarantee the machinery reliability. Traditional fault diagnosis frameworks are commonly under a default assumption: the training and test samples share the similar distribution. However, it is nearly impossible in real industrial applications, where the operating condition always changes over time and the quantity of the same-distribution samples is often not sufficient to build a qualified diagnostic model. Therefore, transfer learning, which possesses the capacity to leverage the knowledge learnt from the massive source data to establish a diagnosis model for the similar but small target data, has shown potential value in machine fault diagnosis with small sample size. In this paper, we propose a novel fault diagnosis framework for the small amount of target data based on transfer learning, using a modified TrAdaBoost algorithm and convolutional neural networks. First, the massive source data with different distributions is added to the target data as the training data. Then, a convolutional neural network is selected as the base learner and the modified TrAdaBoost algorithm is employed for the weight update of each training sample to form a stronger diagnostic model. The whole proposition is experimentally demonstrated and discussed by carrying out the tests of six three-phase induction motors under different operating conditions and fault types. Results show that compared with other methods, the proposed framework can achieve the highest fault diagnostic accuracy with inadequate target data.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.