Domain Generalization (DG) has been explored to achieve machine fault diagnosis under previously unseen operating conditions. However, most DG methods assume access to training data collected across multiple conditions, an assumption that rarely holds in industrial practice, where fault data are typically available from only a single operating condition. To address this critical constraint, we propose an attention-throughout latent diffusion model for single-source domain generalization (ATLD-SSDG). The proposed framework learns discriminative fault representations from a single-condition source domain and generalizes robustly to multiple unseen target conditions. First, to effectively capture complementary fault information, vibration signals from three views are fused and projected into a latent space via a collaborative attention fusion mechanism. Next, a dedicated one-dimensional (1D) U-Net is constructed to address information loss in existing approaches and facilitate more effective conditional diffusion. Unlike existing methods that directly adopt computer vision diffusion architectures, the proposed 1D U-Net is specifically designed for vibration signals, preserving localized fault-related details and preventing information loss caused by time–frequency transformations. Moreover, by explicitly regulating self-attention and cross-attention within the diffusion model, the framework preserves fault-relevant characteristics while selectively substituting operating-condition-related factors, thereby enabling controllable and effective domain generalization. Extensive experiments demonstrate superior generalization performance and diagnostic accuracy of the proposed method over state-of-the-art DG methods. These results indicate that latent diffusion, when properly structured for 1D condition-monitoring signals, provides an effective mechanism for single-source domain generalization, helping to close an important gap in DG research for predictive maintenance.

The core objective of graph neural network (GNN)-based remaining useful life (RUL) prediction methods for equipment with multi-source sensors is to learn effective graph representations, and graph pooling is an efficient approach to achieve it. However, existing graph pooling techniques are limited in modeling hierarchical structures and have limitations in embedding space representation. To overcome these limitations, a dynamic curvature pooling graph convolutional network (DCPGCN) is proposed for RUL prediction of equipment with multi-source sensors. DCPGCN develops a hyperbolic hierarchical graph pooling framework. By leveraging the geometric advantages of hyperbolic space for hierarchical representation, the proposed framework more effectively captures multi-level structural information in graphs, significantly improving the overall structural fidelity of the graph representation. Moreover, a curvature predictor driven by pooling path deviation is proposed. By quantifying the geometric distortion along leaf-to-root paths in hyperbolic space, the predictor dynamically adjusts the curvature parameter, improving the embedding space’s adaptability and expressiveness for the graph’s hierarchical structure. Finally, experiments on the CMAPSS dataset demonstrate that the proposed method outperforms multiple state-of-the-art approaches in prediction accuracy, while experiments on real-world wind turbine RUL prediction further confirm its superiority and potential in engineering applications.