Huasheng Xie

A high Fe content easily produces Fe-rich phases with a harmful morphology, resulting in a huge detrimental effect on the properties and recycling ability of Al-Si alloys. Therefore, finding ways to effectively transform Fe-rich phases to form a beneficial phase or shape is of great significance. Accordingly, Al-Si-based alloys with Fe contents ranging from 0.1 wt.% to 2.0 wt.% were modified by different Mn additions. Moreover, experiments combined with simulations were utilized to comprehensively analyze the mechanism of Mn on the morphology and microstructural evolution of Fe-rich phases from different perspectives. The current findings determine that adding different Fe contents changes the phase-transition reactions in alloys. Without Mn, and by increasing the Fe content from 0.1 wt.% to 2.0 wt.%, the Fe-rich phases gradually convert from a skeleton-shaped α-Al₈Fe₂Si (β-Al₉Fe₂Si₂ with a fibrous (0.5 wt.%), needle-like (1.0 wt.%) and plate-like shape without curvatures (2.0 wt.%). The maximum length and mean aspect ratio increase from 12.01 μm to 655.66 μm and from 1.96 to 84.05, while the mean curvature decreases from 8.66 × 10⁻² μm⁻¹ to 8.25 × 10⁻⁴ μm⁻¹. The addition of 0.35 wt.% Mn promotes a new Chinese-character and petal-shaped α-Al₁₅(FeMn)₃Si₂, with an atomic ratio of Fe and Mn of 1:1 when the Fe content is lower than 0.5 wt.%, while it transforms to β-Al₁₅(FeMn)₃Si₂ with an atomic ratio of 5:1, presenting as a refined plate-like shape with a certain curvature, as the Fe content increases to 2.0 wt.%. Mn alters the phase reactions and increases the threshold of the Fe content required for β-Al₁₅(FeMn)₃Si₂, limiting the formation and growth of them simultaneously in time and space. The enrichment of Mn atoms and solute diffusion at the growth front of β-Al₁₅(FeMn)₃Si₂, as well as the strong atomic-binding ability, can deflect the growth direction of β-Al₁₅(FeMn)₃Si₂ for it to have a certain curvature. Additionally, the enriched Mn atoms easily form α-Al₁₅(FeMn)₃Si₂ and cause the long β-Al₁₅(FeMn)₃Si₂ to be broken and refined to further reduce the damages caused to the alloy’s performance. Ultimately, the maximum length and mean aspect ratio can be effectively reduced to 46.2% and 42.0%, respectively, while the mean curvature can be noticeably increased by 3.27 times with the addition of Mn.

Data imbalance is a crucial factor that limits the performance of automatic defect recognition systems in castings. The bias and deterioration of the model are generated by massive normal samples and minor defect samples. Traditional re-sampling methods randomly change the data distribution and ignore the significant intra-class difference among all normal samples. Therefore, this paper proposes a distribution-preserving under-sampling method for imbalance defect-recognition in castings. In detail, our method divides all normal samples into several sub-groups by cluster analysis and reassembles them into some balance datasets, which makes the normal samples in all balance datasets have an identical distribution with the original imbalance dataset. Finally, experiments on our dataset with 3260 images indicate that the proposed method achieves a 0.816 AUC (area under curve) score, which demonstrates significant advantages compared to cost-sensitive learning and re-sampling methods.

Tailored properties hot forming of 38MnB5Nb by partition cooling can be configured on-demand. A gradient distribution microstructural characteristic can be generated by undertaking an appropriate cooling-tempering process for the regions with different properties requirements before forming. A unified viscoplastic constitutive model coupled with the primary cooling temperature and related material constants is established based on genetic algorithm. Meanwhile, the use of the mixture of jet air and dry ice particles to cool the partition region is essential to achieve different primary cooling temperatures. In this paper, the inverse heat conduction problem is solved to obtain the relationship between the interfacial heat transfer coefficient and different cooling conditions in the partition cooling process. The U-shaped part is taken as an example to simulate the change of temperature, stress-strain, thickness, and spring-back in the process of partition cooling and tailored hot forming properties. The results show that the gradient microstructural characteristic formed by partition cooling has a great influence on the stress field distribution. The maximum stress of 345 MPa can be reached after complete pressure holding in the partition slow cooling tailored properties of the hot forming process. The maximum stress can reach 743 MPa in the partition fast cooling tailored properties of the hot forming process due to the relatively high deformation resistance of the tempered martensite at the bottom center. The maximum residual stress in the two processes after spring-back drops to 305 MPa and 545 MPa, respectively. The spring-back is small under the two processes, with a maximum spring back angle of no more than 1°.

Grain size plays a fundamental role in the mechanical properties of materials. Recently, automatic measurement of average grain size attacks more and more attention based on computer vision. However, low contrast, twin grains, thin boundary, and low connectivity limit the achievement of automatic and accurate image analysis. Inspired by the calculation procedure of grain size, we propose a center-guided and connectivity-preserving network for grain boundaries segmentation. On the one hand, the proposed center feature extraction module and center-guided feature recalibration mechanism make the network pay more attention to the center area. On the other hand, a connectivity-preserving loss function is integrated with the network, which forces the network to converge towards high connectivity. Benefit by above aspects, our network can segment the grain boundary with high structural integrity and avoid the complex post-processing process. Experiments on the SRIF-GSM dataset reveal that our method achieves 85.98 mIoU and 95.60 clDice scores, demonstrating significant advantages compared with the state-of-the-art semantic segmentation methods.