Abstract:The rapid expansion of data has propelled the progress of deep learning-oriented monitoring of the well-being of industrial machinery. Because of the variance in distributions within health surveillance data acquired from distinct operational circumstances or devices, models trained on health monitoring data from a specific scenario or device might exhibit inadequate performance in real-world implementations. Domain adaptive technology has been widely explored and successfully applied in fault diagnosis, aiming to solve the problem of unlabeled target domain data and a certain distribution deviation from the source domain data. It has the capability to transfer the acquired knowledge from categorized information to untagged data and achieve good diagnostic results. Most existing domain adaptation-based methods typically use adversarial learning to extract domain invariant feature. However, in the process of adversarial learning, the obtained features may exhibit a partial reduction in differentiating details to a certain degree, which limits the further improvement of diagnostic model performance. Therefore, this paper proposes an adversarial-based consistency learning method(ACL) that focuses on improving the discriminability of features while obtaining domain invariant feature representations. Specifically, the proposed method includes a feature extractor, a domain discriminator, and two health status classifiers. In the training process, the domain invariant feature is obtained based on adversarial learning, and the identification of features is enhanced by minimizing the triple loss function. In addition, for unlabeled target domain data, the ability of the model to extract robust features is enhanced through category consistency learning of two health state classifiers. Finally, the efficacy of the method introduced has been showcased through experimentation.

Adversarial Domain Adaptation with Classifier Alignment for Cross-Domain Intelligent Fault Diagnosis of Multiple Source Domains

Adversarial Domain Adaptation With Dual Auxiliary Classifiers for Cross-Domain Open-Set Intelligent Fault Diagnosis

Adversarial multi-domain adaptation for machine fault diagnosis with variable working conditions

Joint Discriminative Adversarial Domain Adaptation for Cross-Domain Fault Diagnosis

Double-level Adversarial Domain Adaptation Network for Intelligent Fault Diagnosis.

A Hybrid Adversarial Domain Adaptation Network for Bearing Fault Diagnosis Under Varying Working Conditions

Multi-source Alignment Domain Adaptation with Similarity Measurement for Cross-domain Bearing Fault Diagnosis

Dual-weight Attention-based Multi-source Multi-stage Alignment Domain Adaptation for Industrial Fault Diagnosis

Domain Adaptation With Multi-Adversarial Learning for Open-Set Cross-Domain Intelligent Bearing Fault Diagnosis

Adversarial Consistency Learning for Domain Adaptation Intelligent Fault Diagnosis

Duplex Adversarial Domain Discriminative Network for Cross-Domain Partial Transfer Fault Diagnosis

Single Source Cross-Domain Bearing Fault Diagnosis Via Multi-Pseudo Domain Augmented Adversarial Domain-Invariant Learning

Multisource Domain Feature Adaptation Network for Bearing Fault Diagnosis Under Time-Varying Working Conditions

Adversarial Multiple-Target Domain Adaptation for Fault Classification

Deep Adversarial Subdomain Adaptation Network for Intelligent Fault Diagnosis

A Dual-View Alignment-Based Domain Adaptation Network for Fault Diagnosis

A New Multisource Domain Bearing Fault Diagnosis Method With Adaptive Dual-Domain Obfuscation Weighting Strategy

Multi-source Unsupervised Domain Adaptation for Machinery Fault Diagnosis under Different Working Conditions

A new multi-source information domain adaption network based on domain attributes and features transfer for cross-domain fault diagnosis

Pair-Wise Orthogonal Classifier Based Domain Adaptation Network for Fault Diagnosis in Rotating Machinery

Open-set Federated Adversarial Domain Adaptation Based Cross-Domain Fault Diagnosis