|
|
|
| Fault Diagnosis Method Based on Multilayer Hypergraph Convolutional Neural Network |
| ZHANG Yuandong1, ZHANG Xianjie2, ZHANG Ruonan1, ZHANG Haifeng2
|
1. East China Electric Power Test and Research Institute, China Datang Corporation Science and Technology Research Institute Co, Ltd, Hefei 230000, China;
2. School of Mathematical Sciences, Anhui University, Hefei 230601, China |
|
|
|
|
Abstract Machine learning methods have made significant advancements in the field of fault diagnosis, especially for the complex industrial processes. However, most existing methods only consider the features of individual samples or pairwise relationships between samples, rarely taking into account higher-order relationships among samples and structural diversity among samples. Therefore, this paper proposes a fault diagnosis method based on a multilayer hypergraph convolutional neural network. The method first utilizes multiple similarity indicators to construct multilayer hypergraphs with different structures. Then, it performs intralayer hypergraph convolution and interlayer graph convolution operations to extract and fuse features. Experiments are conducted on the simulation dataset of SEU and the real dataset of the coal mill unit, and the results show that this method can effectively improve the accuracy of fault diagnosis.
|
|
Received: 11 August 2023
Published: 27 April 2025
|
|
|
|
|
|
|
|
|
