To address the critical challenges of submerged weak fault signatures and extreme operational variability in aerospace bearing diagnosis, a GADF-powered dual-branch network with KAN-Swin Transformer is proposed for fault diagnosis of aerospace bearing in this study. Firstly, the one-dimensional vibration signal is encoded into a two-dimensional time-frequency image through gramian angular difference field (GADF), employing polar coordinate mapping to preserve temporal dependencies and spectral dynamic characteristics while addressing the noise sensitivity limitations of conventional time-frequency analysis methods. Subsequently, a KAN-Swin Transformer module is developed by replacing traditional multilayer perceptron with B-spline basis functions, which enhances nonlinear mapping capability through dynamic grid adjustment strategy, effectively reducing parameter complexity while improving modeling of transient impacts and periodic patterns. Furthermore, a dual-branch parallel architecture is proposed: The KAN-Swin Transformer branch extracts local structural features through hierarchical window attention mechanisms, while the CNN-GAM branch strengthens global texture perception via multi-scale convolution integrated with channel-spatial attention fusion. Finally, cross-modal feature concatenation is combined with adaptive pooling to achieve synergistic optimization of global-local characteristics, significantly enhancing fault pattern discriminability under complex noise environments. The developed method tested on two different sets of aerospace bearing data, has achieved a classification accuracy of 100%. Meanwhile, the collaborative effect of module integration including KAN, Swin Transformer and CNN-GAM is validated through ablation experiments, showing enhanced cross-speed operational recognition rates compared to baseline models and confirming the robustness of the framework against noise and variable loading conditions. By synergistically integrating mechanism fusion and attention architectures, the proposed framework provides a reliable solution for intelligent health monitoring of aerospace bearings.

Ultra-high-definition (UHD) image restoration is vital for applications demanding exceptional visual fidelity, yet existing methods often face a trade-off between restoration quality and efficiency, limiting their practical deployment. In this paper, we propose TSFormer, an all-in-one framework that integrates Trusted learning with Sparsification to boost both generalization capability and computational efficiency in UHD image restoration. The key to sparsification is that only a small amount of token movement is allowed within the model. To efficiently filter tokens, we use Min- $p$ with random matrix theory to quantify the uncertainty of tokens (lower trustworthiness), thereby improving the robustness of the model. Our model can run a 4K ( $3840\times 2160$ ) image in real time (40fps) with 3.38 M parameters. Extensive experiments demonstrate that TSFormer achieves state-of-the-art restoration quality while enhancing generalization and reducing computational demands. In addition, our token filtering method can be applied to other image restoration models to effectively accelerate inference and maintain performance.