基于流的遥感图像超分辨率重建

为了提高遥感图像超分辨率重建的质量，提出了一种基于流的遥感图像重建算法。首先，在Glow模型的基础上引入改进后的RRDB架构用于低分辨率图像特征提取，通过构建更多层和连接以提升训练的稳定性。然后，以一种纯数据驱动的流模型来训练分布的参数，通过最大化负的对数似然的方法进行优化，得到该算法的损失函数。实验证明该模型在网络训练过程中能够快速达到稳定收敛的状态，且具有很强的泛化能力。用重建出的图像质量对比SRCNN、SRGAN、ESRGAN，经过测试后发现，提出的算法远远优于SRCNN算法，与其他算法相比也有明显优势。重建出的图像不仅在指标上有所提升，例如与SRCNN相比，PSRN和SSIM分别提升了15%和40%，且人眼观察时有更高的清晰度，高频细节更为丰富。     

Experimental study on flow excursion instability of supercritical hydrocarbon fuel in scramjet regenerative cooling parallel channels

Flow instability of supercritical hydrocarbon fuel is a crucial issue in scramjet regenerative cooling structure. In this study, flow excursion instability and flow distribution in parallel tubes were experimentally studied for supercritical fluids. Two types of flow excursion occur in a single tube. Type I and Type II excursions, and they are corresponding to decreasing and increasing flow rate respectively. They can trigger flow maldistribution between parallel tubes and the hysteresis phenomenon of flow distribution. The effects of system parameters, including inlet temperature, system pressure, and heat flux, on flow distribution were analyzed. In addition, the relationship between flow excursion and the pseudo-critical interval proposed in the literature was established according to the heated tube outlet temperature at the onset of flow instability. Finally, the flow excursion instability boundary was obtained using two dimensionless parameters. These experimental results can provide helpful insight on the mechanism of Scramjet regenerative cooling.     

Trailing-edge shock loss control with self-sustaining synthetic jet in a supersonic compressor cascade

To effectively reduce the loss of strong shock wave at the trailing edge of the supersonic cascade under high backpressure, a shock wave control method based on self-sustaining synthetic jet was proposed. The self-sustaining synthetic jet was applied on the pressure side of the blade with the blow slot and the bleed slot arranged upstream and downstream of the trailing-edge shock,respectively. The flow control mechanism and effects of parameters were investigated by numerical simulation. The res...     

边缘网络下多无人机协同计算和资源分配联合优化策略

针对边缘网络环境下多人机之间存在计算负载不均，造成卸载任务失败的问题，提出了一种多无人机间协作的智能任务卸载方案。通过联合考虑多无人机任务分配、计算资源分配和无人机飞行轨迹，引入公平性指数建立了无人机公平负载最大化和能量消耗最小化问题。基于多智能体深度强化学习框架，提出了融合轨迹规划和任务卸载的分布式算法。仿真结果表明，所提出的多无人机协作方案可以显著提高任务完成率和负载公平度，并且有效适用于大规模用户设备场景。     

Flow field design and experimental investigation of electrochemical machining on blisk

Flow field is a crucial factor to influence the stability and surface quality in the electrochemical machining (ECM) of blisks. A four-way flow mode was proposed to eliminate mixing regions of electrolyte at the leading and tailing edges. Two flow field models were described separately in this report: a W-shaped flow mode and a four-way flow mode. The flow field was analyzed through a finite element method. The results showed that, in comparison with the W-shaped flow mode, the distribution of electrolyte flow was more uniformed and the mixed region in the flow channel was improved. The pressure of the leading and tailing edges inlets was optimized, and optimal pressure of 0.6 MPa was determined. In addition, verification experiments were performed, and the results showed that the stability, efficiency, and quality of the profiles of the blisk blade manufactured by ECM were enhanced in the new flow mode.     

Optimization of flow field in electrochemical trepanning of integral cascades (Ti6Al4V)

Ti6Al4V is widely applied in the integral cascades of aero engines. As an effective machining method, electrochemical trepanning (ECTr) has unique advantages in processing surface parts made of hard-to-cut materials. In ECTr, the state of the flow field has a significant effect on processing stability and machining quality. To improve the uniformity of the flow field when ECTr is applied to Ti6Al4V, two different flow modes are designed, namely full-profile electrolyte supply (FPES) and edges electrolyte supply (EES). Different from the traditional forward flow mode, the flow directions of the electrolyte in the proposed modes are controlled by inlet channels. Simulations show that the flow field under EES is more uniform than that under FPES. To further enhance the uniformity of the flow field, the structure of EES is optimized by modifying the insulating sleeve. In the optimized configuration, the longitudinal distance between the center of the inlet hole and the center of the blade is 6.0 mm, the lateral distance between the centers of the inlet holes on both sides is 16.5 mm, the length to which the electrolyte enters the machining area is 1.5 mm, and the height of the insulating sleeve is 13.5 mm. A series of ECTr experiments are performed under the two flow modes. Compared with EES, the blade machined by FPES is less accurate and has poorer surface quality, with a surface roughness (R_a) of 3.346 μm. Under the optimized EES, the machining quality is effectively enhanced, with the surface quality improved from R_a = 2.621 μm to R_a = 1.815 μm, thus confirming the efficacy of the proposed methods.     

Recent progress of machine learning in flow modeling and active flow control

In terms of multiple temporal and spatial scales, massive data from experiments, flow field measurements, and high-fidelity numerical simulations have greatly promoted the rapid development of fluid mechanics. Machine Learning(ML) provides a wealth of analysis methods to extract potential information from a large amount of data for in-depth understanding of the underlying flow mechanism or for further applications. Furthermore, machine learning algorithms can enhance flow information and automat...     

Sequential dynamic resource allocation in multi-beam satellite systems:A learning-based optimization method

Multi-beam antenna and beam hopping technologies are an effective solution for scarce satellite frequency resources. One of the primary challenges accompanying with Multi-Beam Satellites(MBS) is an efficient Dynamic Resource Allocation(DRA) strategy. This paper presents a learning-based Hybrid-Action Deep Q-Network(HADQN) algorithm to address the sequential decision-making optimization problem in DRA. By using a parameterized hybrid action space,HADQN makes it possible to schedule the beam patte...     

Time-resolved visualization of coherent structures during supersonic boundary layer transition

The coherent structures arising during flat-plate boundary layer transition at Mach number 3.4 are investigated using a custom-built hyper-rate imaging system. The evolution of transitional structures is investigated in the Eulerian and Lagrangian reference frames. The upstream evolution of transition is dominated by the generation of new hairpin structures, while the interaction among multiple structural types dominates the evolution downstream. The breakdown of the existing structure, which may be caused by interactions among multiple types of structures with similar scales, is also visualized.