Xu Shen

Graphene is considered one of the most promising flexible transparent electrode materials as it has high charge carrier mobility, high electrical conductivity, low optical absorption, excellent mechanical strength, and good bendability. However, graphene-based flexible transparent electrodes face a critical challenge in balancing electrical conductivity and optical transmittance. Here, we present a green and scalable direct ink writing (DIW) strategy to fabricate graphene grid patterns by optimizing ink formulation with sodium dodecyl sulfate (SDS) and ethanol. SDS eliminates the coffee ring effect via Marangoni flow, while ethanol enhances graphene flake alignment during hot-pressing, achieving a high conductivity of 5.22 × 10⁵ S m⁻¹. The grid-patterned graphene-based flexible transparent electrodes exhibit a low sheet resistance of 21.3 Ω/sq with 68.5% transmittance as well as a high stability in high-temperature and corrosive environments, surpassing most metal/graphene composites. This method avoids toxic solvents and high-temperature treatments, demonstrating excellent stability in harsh environments.

In this paper, the physical simulation of filling process of vertical centrifugal casting (VCC) of complex titanium alloy casting was studied. Combined with the mature PTV particle tracking technology, the high-speed photography pictures of the filling process of VCC at different rotational speeds were obtained. The trajectory and velocity information of tracer particles in the rotating flow field were obtained by the corresponding analysis software. Then, through the analysis and modeling of quantitative experimental data, the flow behavior characteristics and movement law of titanium alloy melt in the mold cavity under different mold speeds were studied. The results show that: 1. When the mold is still, the front edge of the filling fluid forms a curved surface with the curvature center pointing to the outside of the mold; when the mold rotates, the front edge of the liquid flow forms a curved surface with the curvature center pointing to the inside of the mold; 2. With the increase of the mold rotation speed, the speed of the fluid filling the mold increases significantly; when the rotational speed is greater than 120 rpm, the fluid still has a certain driving force in the mold center far away from the gate It is good for filling the inner corner of mold with fluid; 3. When the rotational speed of centrifugal casting of titanium alloy reaches 45 rpm or above, typical turbulent vortices appear in the wake; with the increase of rotating speed to 180 rpm, the average curvature radius of turbulent vortices first increases and then decreases, and reaches the minimum value of 0.67 cm at 120 rpm.

Knowledge Distillation (KD) is a popular technique to transfer knowledge from a teacher model or ensemble to a student model. Its success is generally attributed to the privileged information on similarities/consistency between the class distributions or intermediate feature representations of the teacher model and the student model. However, directly pushing the student model to mimic the probabilities/features of the teacher model to a large extent limits the student model in learning undiscovered knowledge/features. In this paper, we propose a novel inheritance and exploration knowledge distillation framework (IE-KD), in which a student model is split into two parts - inheritance and exploration. The inheritance part is learned with a similarity loss to transfer the existing learned knowledge from the teacher model to the student model, while the exploration part is encouraged to learn representations different from the inherited ones with a dis-similarity loss. Our IE-KD framework is generic and can be easily combined with existing distillation or mutual learning methods for training deep neural networks. Extensive experiments demonstrate that these two parts can jointly push the student model to learn more diversified and effective representations, and our IE-KD can be a general technique to improve the student network to achieve SOTA performance. Furthermore, by applying our IE-KD to the training of two networks, the performance of both can be improved w.r.t. deep mutual learning. The code and models of IE-KD will be make publicly available at https://github.com/yellowtownhz/IE-KD.

Purpose: The purpose of this study is to simulate the temperature distribution during an electron beam freeform fabrication (EBF³) process based on a fully threaded tree (FTT) technique in various scales and to analyze the temperature variation with time in different regions of the part. Design/methodology/approach: This study presented a revised model for the temperature simulation in the EBF³ process. The FTT technique was then adopted as an adaptive grid strategy in the simulation. Based on the simulation results, an analysis regarding the temperature distribution of a circular deposit and substrate was performed. Findings: The FTT technique was successfully adopted in the simulation of the temperature field during the EBF³ process. The temperature bands and oscillating temperature curves appeared in the deposit and substrate. Originality/value: The FTT technique was introduced into the numerical simulation of an additive manufacturing process. The efficiency of the process was improved, and the FTT technique was convenient for the 3D simulations and multi-pass deposits.

The mesoscopic grain model is a multiscale model which takes into account both the dendrite growth mechanism and the vast numerical computation of the actual castings. Due to the pursuit of efficient computation, the mesoscopic grain calculation accuracy is lower than that of dendrite growth model. Improving the accuracy of mesoscopic grain model is a problem to be solved urgently. In this study, referring to the calculation method of solid fraction in microscopic dendrite growth model, a cellular automata model of 3D mesoscopic grain evolution for solid fraction calculated quantitatively at the scale of cell is developed. The developed model and algorithm validation for grain growth simulation is made by comparing the numerical results with the benchmark experimental data and the analytical predictions. The results show that the 3D grain envelopes simulated by the developed model and algorithm are coincident with the shape predicted by the analytical model to a certain extent. Then, the developed model is applied to the numerical simulation of solidification process of nickel-based superalloys, including equiaxed and columnar dendritic grain growth. Our results show good agreement with the related literature.