Yi Zhang

Condition monitoring of modern industrial machinery increasingly relies on data-driven deep learning methods. However, domain shifts caused by dynamic working conditions often lead to substantial accuracy degradation in unseen target domains. Single-modality sensing also suffers from inherent perception limitations under heavy interference. To address these issues, this paper proposes a multi-modal domain generalization framework named the uncertainty-aware cross-modal fusion network (UCFNet). The framework incorporates three key modules. Specifically, a parallel mixed attention strategy captures fine-grained cross-modal interactions and intra-modal context, thereby extracting domain-invariant representations. In addition, a dynamic collaborative gating mechanism filters out feature redundancy and suppresses interference caused by working condition shifts. Furthermore, an uncertainty-aware feature fusion module estimates the prediction confidence of each modality and uses it to guide decision fusion in unseen domains. Comprehensive experiments are conducted on reciprocating compressor and wheelset bearing datasets. Experimental tasks evaluate both multi-source domain generalization and single-source domain generalization. The results demonstrate that UCFNet achieves superior performance compared with state-of-the-art methods.

Machinery typically operates under constantly changing working conditions in real-time production. Directly applying a diagnostic model, which has been trained solely on monitoring data from a single working condition, to a new unseen working condition poses a notably challenging task called single-source domain generalization (SSDG) fault diagnosis. Current research mainly aims to augment source-domain training samples but struggles to maintain health state information invariance for diverse augmented samples. Therefore, a targeted augmentation domain-mixed network (TADNet) is proposed in this paper for SSDG fault diagnosis of rotating machinery. The TADNet constructs a targeted augmentation chain, which generates samples in an augmented domain with different distributions from the source-domain samples through a chained generation structure, and ensures the semantic consistency of state features. Additionally, a domain random mixing strategy is established to synthesize a new mixed domain with diverse distributions, through probabilistically and randomly mixing the feature statistics of the samples from the source and augmented domains, for further augmentation of samples. The TADNet effectively balances the distribution diversity and semantic consistency of augmented samples in the course of model training. The superior diagnosis generalization ability of the proposed method to unseen working conditions is demonstrated on two rotating machinery datasets.

Single-pixel imaging (SPI), generally based on computational imaging, has the advantages of a wide bandwidth and imaging objects beyond the visual field. However, for general SPI, to obtain a picture with few shadows by one photodetector is a great challenge. To reduce shadows in SPI, this paper proposes a coaxial SPI to reduce shadows in reconstructed images effectively with a simple structure. Unlike the general shadow reduction method in SPI based on image post-processing, the method proposed in this paper is without the need of additional image post-processing. Shadows can be reduced effectively by single view of only one photodetector. The experiments show that the reconstructed image in the coaxial structure is free from shadows, while the picture reconstructed by the non-coaxial structure contains lots of shadows. Moreover, the coaxial structure shows good applicability for the imaging of the end face in a deep hole. The proposed coaxial SPI has provided a potential direction for the applications of the SPI in information extraction and image quality improvement since the shadows in the reconstructed images have been reduced effectively.

This study designed Al_xCrFeNi (x = 0.8, 1.0, 1.2) medium-entropy alloys featuring a BCC + B2 dual-phase structure to systematically investigate the effects of Al content variation and heat treatment on microstructure evolution and corrosion behavior. Microstructural characterization revealed that all investigated alloys maintained the BCC + B2 dual-phase labyrinth structure. Electrochemical tests showed that as the Al content increased, the corrosion current density and corrosion rate in a 3.5 wt% NaCl solution increased. Synergistic analysis of post-corrosion morphology (through electrochemical testing and in-situ immersion) combined with XPS analysis of the passive films revealed that the initial stage of corrosion was primarily pitting. Subsequently, due to the loose and porous Al₂O₃ passive layer formed by the NiAl-rich phase, which was easily attacked by Cl⁻ ions, the corrosion progressed into selective corrosion of the NiAl phase. Notably, heat treatment at 1000 °C induced microstructural refinement with enhanced coupling between chunky and labyrinth structures, resulting in improved corrosion resistance despite a 4–6% reduction in Vickers hardness due to elemental homogenization. Among the investigated alloys, the heat-treated Al_0.8CrFeNi exhibited the most promising corrosion resistance.

At the initial rolling temperature of 400 °C, ZK60 magnesium alloy was hot rolled by three different rolling paths with different roll speed ratios (RSR) of 1:1.15, 1:1.2, and 1:1.5, respectively. The effects of different rolling processes on the microstructure and mechanical properties of the alloy were studied. The microstructure, plasticity, strength, hardness, and texture intensity of rolled samples were analyzed in this work. The results show that the microstructure uniformity of the alloy under multi-path asynchronous rolling (MAR) is significantly improved, which improves the mechanical properties of the material to a certain extent, and effectively weakens the texture intensity of the basal plane and reduces the anisotropy. The amount of randomly oriented grains of ZK60 magnesium alloy rolled by the C-1.5 (path C combined with the RSR of 1:1.5) process are significantly increased, which significantly weakens the basal texture and improves the ductility of the alloy, greatly enhancing the processing and formability of ZK60 magnesium alloy.

The use of zinc (Zn) alloys as a biodegradable metal for medical purposes has been a popular research topic. This study investigated the strengthening mechanism of Zn alloys to enhance their mechanical properties. Three Zn-0.45Li (wt.%) alloys with different deformation amounts were prepared by rotary forging deformation. Their mechanical properties and microstructures were tested. A simultaneous increase in strength and ductility was observed in the Zn-0.45Li alloys. Grain refinement occurred when the rotary forging deformation reached 75.7%. The surface average grain size reached 1.19 ± 0.31 μm, and the grain size was uniformly distributed. Meanwhile, the maximum elongation of the deformed Zn-0.45Li was 139.2 ± 18.6%, and the ultimate tensile strength reached 426.1 ± 4.7 MPa. In situ tensile tests showed that the reinforced alloys still broke from the grain boundary. Continuous and discontinuous dynamic recrystallization during severe plastic deformation produced many recrystallized grains. During deformation, the dislocation density of the alloy first increased and then decreased, and the texture strength of the (0001) direction increased with deformation. Analysis of the mechanism of alloy strengthening showed that the strength and plasticity enhancement of Zn-Li alloys after macro deformation was a combination of dislocation strengthening, weave strengthening, and grain refinement rather than only fine-grain strengthening as observed in conventional macro-deformed Zn alloys.

The metallization of polyetherimide (PEI) is widely considered to enhance its surface properties and enhance its application in engineering fields; however, adhesion is a key factor in determining the reliability of PEI metallization. A Ni/Cu/Ni multilayer coating was successfully manufactured on a batch of PEI resin reinforced with glass fibers by a two-step metallization process, including sandblasting and activation/acceleration. The microstructure and morphology of the top-surface and cross-section of the coatings were observed by scanning electron microscopy. The chemical state and composition of the deposits were characterized by both X-ray photoelectron and energy-dispersive spectroscopy. The adhesion state was qualitatively evaluated by cross-cut tests with 3M tape. The surface roughness of the substrate significantly increased after the sandblasting process, which could improve the adhesion between the multilayer coating and the PEI substrate. After the standard activation process, the acceleration made an effect on the deposition of the initial Ni layer for electroless plating. The influence of acceleration on the appearance quality of metallization on the PEI substrate was studied and, at the same time, the mechanism of acceleration was investigated and addressed.