Xi’an University

Cu-Sn alloy is a key raw material used in the preparation of Nb3Sn superconducting wires using the bronze method. The mechanical properties of this Cu-Sn alloy are directly responsible for determining the properties of the superconducting wires. The superconducting properties of Nb3Sn can be significantly improved by Ga doping. However, the effect of Ga doping and the amount thereof on the properties of Cu-Sn alloys has rarely been reported. In this study, the distribution of Ga and its impact on the mechanical properties of Cu-Sn alloys are investigated via experiments and simulations. The results indicate that Cu and Ga readily produce strong electron exchange and a stable solid-solution structure, leading to Ga superiority in the solid-solution competition with Sn and to δ phase segregation. For the first time, a 1.0 wt.% Ga-added Cu-Sn alloy exhibiting an excellent elongation of 100.8% was successfully prepared. Further investigation revealed that with increasing Ga content, the activation of planar fault slip systems becomes more difficult, the volume fraction of planar fault structures gradually decreases, and the dislocation dissociation distance decreases as cross-slip occurs. The simulation results revealed that as Ga doping increases, the matrix stacking fault energy increases and the deformation mechanism shifts from being dominated by twinning to a balanced combination of slip and twinning, which is the primary mechanism for the synergistic enhancement of strength and ductility in the alloy. In addition, Ga doping significantly elevated the superconducting transition temperature of Nb3Sn by approximately 0.85 K and improved the critical current density. This study innovatively demonstrates that compared to the front-end Cu-Sn alloy, the addition of 1.0 wt.% Ga achieves synergistic property enhancement of the back-end Nb3Sn superconducting wire. This study provides both a theoretical and experimental foundation for the preparation of Cu-Sn alloys with elevated properties for use in Nb3Sn superconducting wires.

To address challenges related to complex background interference and insufficient multi-scale target feature extraction in lettuce growth stage detection. The lightweight YOLOv10 detection model and the specific characteristics of lettuce field data were used. The CNCM channel non-local mixture mechanism and ASF adaptive spatial frequency attention mechanism were incorporated to optimize lightweight modules, including DownSample, Zoom_cat, and ScalSeq, within the original model. Consequently, an improved CCASF-YOLOv10 model was constructed, integrating multi-scale feature fusion and enhanced target feature extraction. Experimental results demonstrate that, in an NVIDIA A40 GPU testing environment, the model achieves an accuracy rate of 91.9%, a recall rate of 91.6%, mAP@0.5 of 95.3%, and mAP@0.5:0.95 of 72.9%. The parameter size is 11.9 M, and the single-frame inference speed is 24.76 ms, indicating a favorable balance between detection precision, model efficiency, and real-time inference. Furthermore, an intelligent machine vision detection system for lettuce growth-stage monitoring and precise field control was developed using the CCASF-YOLOv10 model. This system facilitates the industrial advancement of lettuce cultivation.

Multimodal large language models (MLLMs), which integrate language and visual cues for problem-solving, are crucial for advancing artificial general intelligence (AGI). However, current benchmarks for measuring the intelligence of MLLMs suffer from limited scale, narrow coverage, and unstructured knowledge, offering only static and undifferentiated evaluations. To bridge this gap, we introduce MDK12-Bench, a large-scale multidisciplinary benchmark built from real-world K–12 exams spanning six disciplines with 141K instances and 6,225 knowledge points organized in a six-layer taxonomy. Covering five question formats with difficulty and year annotations, it enables comprehensive evaluation to capture the extent to which MLLMs perform over four dimensions: 1) difficulty levels, 2) temporal (cross-year) shifts, 3) contextual shifts, and 4) knowledge-driven reasoning. We propose a novel dynamic evaluation framework that introduces unfamiliar visual, textual, and question form shifts to challenge model generalization while improving benchmark objectivity and longevity by mitigating data contamination. We further evaluate knowledge-point reference-augmented generation (KP-RAG) to examine the role of knowledge in reasoning. Key findings reveal limitations in current MLLMs in multiple aspects and provide guidance for enhancing model reasoning, robustness, and AI-assisted education.