Sensors ensure the normal operation of the system by monitoring and collecting large amounts of data in real time. Due to the complexity and variability of the working environment, the industrial system exhibits different behaviors under different working conditions. Existing research methods mainly rely on graph structure to learn one-to-one connections between samples, and there is a lack of research on higher-order features and contextual data. And changes in the current moment of time series data may have the same impact on the next moment and some time in the future. Therefore, we propose a multi-channel variational hypergraph auto-encoder network(MC-VHAE) for unsupervised domain adaptation fault diagnosis across working conditions. In the proposed multi-channel network, on the one hand, convolutional neural networks are used to extract multi-scale features from time-domain fault data. On the other hand, discrete wavelet transform is used to separate the high-frequency and low-frequency components of the fault sample in the time-frequency domain. The potential higher order information is extracted by variational hypergraph auto-encoder (VHAE). In VHAE, a multi-order neighborhood hypergraph convolutional layer (MON-HGCL) is designed to aggregate the high-order feature information of different order neighborhoods in the hypergraph nodes. Finally, the feature fusion layer is used to retain the low-frequency trend component and multi-scale features while removing high-frequency noise components. Experimental results show that MC-VHAE outperforms existing methods and demonstrates its ability to extract domain-invariant features under different operating conditions.
Recently, graph contrastive learning (GCL) has achieved remarkable performance in graph representation learning. However, existing GCL methods usually follow a dual-channel encoder network (i.e., Siamese networks), which adds to the complexity of the network architecture. Additionally, these methods overly depend on varied data augmentation techniques, corrupting graph information. Furthermore, they are heavily reliant on large quantities of negative nodes for each object node, which requires tremendous memory costs. To address these issues, we propose a novel and simple graph representation learning framework, named SimGRL. Firstly, our proposed network architecture only contains one encoder based on a graph neural network instead of a dual-channel encoder, which simplifies the network architecture. Then we introduce a distributor to generate triplets to obtain the contrastive views between nodes and their neighbors, avoiding the need for data augmentations. Finally, we design a triplet loss based on adjacency information in graphs that utilizes only one negative node for each object node, reducing memory overhead significantly. Extensive experiments demonstrate that SimGRL achieves competitive performance on node classification and graph classification tasks, especially in terms of running time and memory overhead.
29 March 2025 in Biomedical Signal Processing and Control