山西职业技术学院

Direct conversion of coal-derived syngas to ethanol represents a sustainable route to relieve the dependence on grain fermentation and petroleum-based ethylene hydration, yet catalyst development remains the critical challenge. In this study, binary ZnAl catalysts with varying Zn/Al molar ratios were fabricated through a fully liquid-phase method, and their structure–performance relationship for CO hydrogenation was systematically explored. The Zn/Al ratio effectively modulates catalyst phase composition, specific surface area, surface acidity, oxygen vacancy concentration, and surface chemical states. Notably, the catalyst with Zn/Al = 1.25 exhibits the optimal catalytic performance under mild conditions (280 °C, 4.0 MPa), achieving 4.8% CO conversion, 21.3% total alcohol selectivity (C-mol%), and ethanol proportion of 62.15% in total alcohols. This performance is distinctly superior to most reported Zn-based catalysts under milder reaction parameters. The rapid deactivation of the Zn/Al = 0.5 catalyst is ascribed to the reduction of surface hydroxyls, the weakened CO adsorption capacity, and the suppressed formation of formate and methyl intermediates. Based on these findings, a feasible reaction pathway involving CO adsorption, intermediate conversion, and C–C coupling to ethanol is proposed. This work offers a valuable reference for constructing non-Cu ZnAl catalysts with high efficiency for syngas-to-ethanol conversion.

Community detection is a fundamental problem in network learning. Existing semi-supervised approaches often underutilize labeled data and semantic structures, and their performance drops significantly when labeled samples are scarce. Many methods rely on known samples to learn community structures, but structural differences can cause mismatches with target communities, reducing detection quality. Ignoring community structure while neglecting user feature information remains a major challenge for improving community detection accuracy. Motivated by the need to effectively integrate network structure and node features, this study proposes a semi-supervised and semi-local community detection model based on a graph convolutional neural network (SLC-GCN). Our approach extracts a semi-local subgraph around the target node and leverages this subgraph to identify communities with similar structural patterns. SLC-GCN simultaneously utilizes both topological structure and user feature information and employs an ensemble clustering approach to integrate the results into a coherent community structure. Furthermore, we extend the identified communities for link prediction in multilayer networks. For this purpose, we introduce a novel similarity metric based on local random walks and reliable weighted paths, which effectively addresses the link prediction problem in multilayer networks while incorporating community information. The effectiveness of SLC-GCN is evaluated on both real and synthetic multilayer networks for the joint community detection and link prediction task. Comparative evaluations reveal that SLC-GCN achieves superior results relative to current leading methods on a wide range of performance metrics.

Dissimilar metal beam-column welds used in earthquake-resistant steel frames often fail prematurely because welding parameters simultaneously govern thermal severity, microstructure gradients, and residual-stress hotspots at constrained root regions. Here, an Integrated Computational Materials Engineering (ICME) digital thread is developed for hybrid laser-arc welding (HLAW) of a dissimilar beam-column joint, linking process inputs to lifecycle performance through validated thermo-mechanical simulation and targeted experiments. A moving hybrid heat-source finite-element model predicts peak temperature and cooling severity (t8/5)and resolves longitudinal residual-stress localization, and is calibrated against thermocouple histories and fusion-profile macrographs. These thermal outputs drive microstructure/interface assessment across the two fusion boundaries (I1 and I2), including coarse-grained HAZ evolution and a graded diffusion/reaction layer at the dissimilar interface, supported by EBSD/EDS and hardness mapping. Property heterogeneity is condensed into damagedriver indices based on HAZ softening ( Delta HVsoft)and hardness-gradient severity (GHV), enabling mechanistic linkage to crack initiation at the root-critical zone (CZ2)under cyclic (seismic-type) loading. Model-guided parameter optimization reduces the dominant damage drivers at CZ2and shifts cyclic damage away from the root, increasing low-cycle fatigue life from similar to 4 & times; 104to similar to 1 & times; 105cycles ( approximate to 2.5 & times; ). Finally, probabilistic propagation of process variability defines a robust operating window that meets a target life with high reliability, providing a process-efficient pathway to qualify high-performance dissimilar joints with reduced experimental trials supported by targeted validation of baseline, optimized conditions and improved lifecycle margins for seismic applications.

The fatigue performance of face gears can be effectively improved through shot peening strengthening. To reveal the influence mechanism of shot peening parameters on the strengthening effect, a random shot peening model was first established, through which the effects of shot peening parameters on the residual stress distribution and tooth surface morphology of face gears were analyzed. Furthermore, a bending fatigue life prediction model considering both residual stress and surface morphology was developed, and the influence mechanisms of process parameters, including shot diameter, shot velocity, and surface coverage, on the bending fatigue life were investigated. The results showed that the compressive residual stresses induced by shot peening could effectively offset the working tensile stresses and reduce the stress peak generated at the tooth root region under external loading conditions. Moreover, compared with surface coverage and shot diameter, shot velocity was found to have a more significant influence on fatigue life, and an appropriate increase in shot velocity could effectively enhance the strengthening effect.

To efficiently and accurately generate ancient architectural components, an algorithm for generating ancient architectural components based on Rhino. This study collects data on ancient architectural components through ground LiDAR field measurements, historical literature mining, and digital model libraries. The data is preprocessed using algorithms such as bilateral filtering and maximum entropy segmentation, and features are extracted using convolutional neural networks. Based on the generative adversarial network framework, targeted improvements are made to its generator and discriminator, incorporating attention mechanisms and size control modules. A multi-scale training strategy and a new loss function are adopted, and Rhino. Inside and Grasshopper are integrated to achieve parameterized design and collaborative generative adversarial network. The results showed that the average peak signal-to-noise ratio of the generated components was 30.2 dB, the mean absolute error of beam size was 8.2 mm, the cosine similarity of decorative patterns was 0.82, and the average time to generate a single component was only 0.7 s. This algorithm outperforms comparative algorithms in visual realism, structural rationality, restoration of historical and cultural features. It has strong applicability to different types of ancient building components, providing strong technical support for the digital protection, restoration, and design of ancient buildings.