Abstract:With the advancement of complex system diagnosis, prediction, and health management technologies, digital twin technology has become a prominent research area in the fields of intelligent manufacturing and system operation and maintenance. However, due to the high complexity of practical systems, the difficulty of data acquisition, and the low accuracy of modeling techniques, current digital twin modeling suffers from low accuracy, and the generalization ability of models is poor when applied in model transfer. To address this issue, a novel fault diagnosis method is proposed, which integrates a digital twin model based on transfer learning. The framework introduces an innovative approach to construct multiple digital twin models using both mechanistic and data-driven models. The mechanism twin constructs a universal simulation model based on physical equipment and updates it with system response measurement data. The data twin consists of a high-dimensional fully connected GAN (HDFC-GAN) twin for extracting deep features from data and an LSTM twin for extracting time series features. Subsequently, transfer learning is introduced to achieve deep fusion in the multiple digital twins system. The mechanism twin is used to obtain source domain samples to construct a diagnostic network, and the data twin is used to extract target domain features to correct the diagnostic network, thereby improving the accuracy and reliability of fault diagnosis. Finally, the proposed framework is applied to the fault diagnosis of triplex pump equipment. The accuracy of diagnosis continuously improves as the system is updated and ultimately reaches 89.28%, demonstrating the effectiveness of the algorithm and providing a novel solution for the generalization limitations of current digital twin models.

An Incremental Model Transfer Method for Complex Process Fault Diagnosis

Fault Diagnosis Method of Multimode Process Based on VMD-IJITL-RLSSVM

Fault Diagnosis in Chemical Processes Based on WPA and WLS-SVM

Multichannel Dynamic Graph Convolutional Network-Based Fault Diagnosis and Its Application in Blast Furnace Ironmaking Process

Deep Learning With Tensor Factorization Layers For Sequential Fault Diagnosis And Industrial Process Monitoring

A New Incremental Training Framework Based on Dynamic Weight Allocation for Intelligent Fault Diagnosis

Process fault diagnosis with model- and knowledge-based approaches: Advances and opportunities

Adaptive Incremental Diagnosis Model for Intelligent Fault Diagnosis with Dynamic Weight Correction

A Feature Extension and Reconstruction Method with Incremental Learning Capabilities under Limited Samples for Intelligent Diagnosis

A New Feature Boosting Based Continual Learning Method for Bearing Fault Diagnosis with Incremental Fault Types

An Imbalance Modified Convolutional Neural Network With Incremental Learning for Chemical Fault Diagnosis

Fault Diagnostic Method Based on Deep Learning and Multimodel Feature Fusion for Complex Industrial Processes

Generalized Transformer in Fault Diagnosis of Tennessee Eastman Process

Latent and Small Fault Detection and Diagnosis for Dynamic Processes

Multiple Local Reconstruction Model-based Fault Diagnosis for Continuous Processes

Fault diagnosis of TE process based on incremental learning

Graph Continual Learning Network: an Incremental Intelligent Diagnosis Method of Machines for New Fault Detection

A Novel Transfer Learning Method for Fault Diagnosis Using Maximum Classifier Discrepancy With Marginal Probability Distribution Adaptation

A Lifelong Learning Method based on Generative Feature Replay for Bearing Diagnosis with Incremental Fault Types

Transfer Learning-based Multiple Digital Twin-assisted Intelligent Mechanical Fault Diagnosis