LAUSR

Conventional intelligent diagnostic model is built on the foundation that the training data and testing data are recorded under the same operating condition, which neglects the fact that the operating condition of the rotating machinery usually varies. The feature distribution of the recorded data in one operating condition may be inconsistent with the feature distribution of the recorded data in another operating condition. It is easy to cause a significant distribution discrepancy between the training data and testing data. To address this issue, an unsupervised domain adaptation approach based on a symmetric co-training framework is proposed in this study. In the proposed symmetric co-training framework, a universal feature extractor and two individual classifiers are built as the main elements. The structures of the two classifiers are symmetric and its parameters are updated in a co-training style. The parameters of the feature extractor and two classifiers are continuously updated via an adversarial training process. The cosine similarity of the predictions from two classifiers is introduced to guide the adversarial training process, which can not only minimize the distribution discrepancies between source domain data and target domain data, but also push the feature subspaces for different healthy conditions away from the class boundaries. The application of the proposed method on two sets of experimental bearing fault data validates that the proposed method can successfully address the domain shift phenomenon between the recorded data under different operating conditions.