一种具有宽带双极化低散射特征的微带阵列天线

针对阵列天线宽带散射缩减设计进行研究,设计了一种基于无源对消技术的低散射阵列天线,该新型微带阵列天线在宽频带内具有双极化低雷达散射截面（RCS, Radar Cross Section）性能;对基于两种散射性能不同的单元组成阵列的RCS性能进行了理论研究,进行了单元的散射幅度和相位对阵列RCS的影响分析;提出了一种加载T型缝隙的新型微带天线结构,该单元结构的辐射性能与散射性能能够进行独立调控和综合优化,该单元与传统微带贴片单元具有相似的辐射特性,并可在宽频带（带内和带外）内与传统微带单元产生有效相位差;将传统微带单元和加载T型缝隙的新型微带单元组成4×4阵列天线,仿真结果表明,提出的阵列天线在3GHz～7GHz（相对带宽80%）频带内实现了同极化RCS缩减,在3.3GHz～7GHz（相对带宽71.8%）频带内实现了交叉极化RCS缩减,缩减峰值分别为16.3dB和36.3dB,带内RCS缩减均值分别为14.1dB 和17.6dB;与传统微带阵列天线相比,提出的阵列天线增益下降小于0.1dB;提出的微带阵列天线具有高效率辐射和宽频带双极化低散射性能,为低散射阵列天线设计提供了新的思路。     

基于 FPGA 的深度可分离卷积加速器研究

设计了一种基于FPGA的低功耗深度可分离卷积加速核;根据PW卷积和DW卷积计算中的共性,采用一种固定乘法阵列通过改变特征和权重输入数据流的方式实现两种卷积的计算结构,最大化DSP的利用率;针对8位非对称量化中符号位可能会溢出的问题,采用符号位单独处理的方法重新封装了双乘法器结构;通过层内7级流水结构保证每个周期数据处理的并行度;在Zynq UltraScale+系列FPGA上成功部署了加速结构;经实验测试,提出的加速结构在提高网络推理速度的同时降低了片上资源的依赖度和整体功耗,原生MobilenetV2在所提FPGA加速器上的平均吞吐率高达130.6GOPS且整体功耗只有4.1w,满足实时边缘计算的要求;相比其他硬件平台,能效比有明显提升;与FPGA上的同类型加速器相比,在性能密度（GOPS/LUT）、功率效率（GOPS/W）和DSP效率（GOPS/DSP）上均有优势。     

基于iSIGHT平台的倾斜冲击射流传热特性优化

为了研究和优化半封闭倾斜射流中击移动平板的传热性能,采用iSIGHT与CFD软件的联合仿真,根据多岛遗传算法动态调整射流角度和平板移动速度,进行狭缝湍流冲击射流的数值模拟.结果表明,平板表面的传热效果主要是由板速决定的,低速时高角度下的平板表面的平均努塞尔数较高;初始设计时流场左侧的回流区和二次回流区消失,优化后的流场结构得到改善,移动平板表面的平均努塞尔数比初始设计结果提高7.62％,热传导更加均匀.     

火箭发动机喷流噪声数值模拟及声振分析

为给火箭系统结构振动响应分析提供有效载荷,采用雷诺平均N S（Reynolds averaged N S,RANS）方程求解喷流流场与用非线性声学求解器（Non linear Acoustics Solver,NLAS）求解喷流声场相结合的方法,对某高超声速火箭液体发动机喷流噪声进行数值模拟.用有限元法和统计能量分析相结合的方法,求解发动机模型在噪声作用下的全频段振动响应.计算结果表明：发动机喷流噪声声压级大小与喷流流场的湍流动能密切相关,湍流强度大的位置喷流噪声声压大;喷流流场初始段混合层内产生的噪声在高频段大于过渡区内产生的噪声,但中低频段却相反.     

CFD并行计算中的多区结构网格二次剖分方法与实现

在多区结构网格计算流体动力学(computational fluid dynamics,CFD)并行模拟中,为了与并行计算资源相适应,经常需要对原始流场网格进行二次剖分与区块分组.在对区块分组和网格二次剖分进行了总结综述的基础上,重点提出针对多区结构网格二次剖分的两种策略:几何剖分和嵌套二分.基于这两种策略完成了剖分软件工具TH-MeshSplit,可实现初级方式、专业方式和专家方式3种运行方式,为用户在自动化与灵活性方面提供了多样化选择.数值实验结果表明,两种剖分策略及其实现软件可在较短时间内完成复杂的剖分,剖分后的网格在负载平衡性、计算通信比等方面具有更优的性能,从而为后续CFD流场的高效并行加速求解奠定了基础.     

电脑文摘

０００６９适用于人工神经网络的双环形收缩阵列网∥ＰｒｏｃｅｅｄｉｎｇｓＣＩＲＣＵＩＴＳ，ＤＥＶＩＣＥＳＡＮＤＳＹＳＴＥＭＳ．—１９９９ ,１４６（５）．—２２５～２２９本文介绍了一种可以用不实现人工神经网络的有效而快速的收缩结构。这种称作ＴＯＲＡＮ（二合一环形阵列网）的结构包含有若干个双环形收缩阵列网络 ,能够并行执行两组加权数据对的乘法运算。与单环形收缩阵列网相比 ,双环形收缩阵列网所需要的硬件虽然增加了４０% ,但它完成加权和计算的循环次数却减少了一半。０００７０交叉稳定式多输出正向伪共振转换器的分…     

微型人造血管壁面特征对血液流动的影响

针对3D打印和喷丝成型两种不同微加工方法制备的人造血管壁面特征对其内部液态流场的影响开展仿真研究。借助商用Comsol软件,研究了不同壁面拓扑形貌,结构尺寸和流体粘度等因素对人造血管内部流场的影响规律分析。结果表明:3D打印制备的人造血管内壁面凸起阵列结构在流场内形成了液体滞留区,该区域面积不随液体粘度的变化而变化,而会随着凸起结构尺寸的变大而变大;当滞留区带有血栓时,滞留区域面积进一步增大,容易引起血小板富集,因此,3D打印的人造血管壁面凸起结构对于血栓具有诱发和强化的作用。     

一种阵列众核处理器的多级指令缓存结构

阵列众核处理器由于其较高的计算性能和能效比已经被广泛应用于高性能计算领域。而要构建未来高性能计算系统处理器必须解决严峻的"访存墙"挑战以及核心协同问题。通常的阵列处理器中,核心多采用单线程结构,以减少开销,但是对访存提出了较高的要求。在阵列众核处理器中,在单核心中引入硬件同时多线程技术,针对实验中一级指令缓存命中率随着线程数增加而显著降低的问题,提出了一种面向阵列众核处理器的冗余指令缓存存储结构,基于该结构,提出采用FIFO及类LRU替换策略。通过上述优化的高速缓存结构设计,经实验模拟,双线程整体指令Cache失效率降低了25.2%,整体CPI性能提升了30.2%。     

可重构结构下合成视点失真变化算法并行设计与实现

针对三维高效视频编码（3D-HEVC）中,基于深度图的合成视点失真变化（SVDC）算法存在计算时间复杂度较高的问题,提出了一种在可重构阵列结构下基于混合粒度的SVDC算法并行化方法。首先,将SVDC算法分为虚拟视点合成（VVS）和失真值计算两个部分。其次,VVS部分采用流水线作业方式加速,而失真值计算部分采用两级划分加速：任务级——将合成后的图像按照像素点进行划分,指令级——将像素点内部的失真值按照计算过程进行划分。最后,采用可重构机制将VVS部分和失真值计算部分进行并行化处理。理论分析和硬件仿真结果表明,在执行时间上,采用4个处理单元（PE）的该方法具有2.11的加速比性能,与基于底层虚拟机（LLVM）和共享存储并行编程（OpenMP）的SVDC算法相比,计算时间分别缩短了18.56%和21.93%。可见所提方法能挖掘SVDC算法的并行性,并结合可重构阵列结构特点有效缩短了SVDC算法的执行时间。     

约束重编程单元数量的忆阻器阵列闭环重映射算法

忆阻器阵列能够有效地加速神经网络中的矩阵运算,但会受到老化的影响,导致忆阻器阵列计算精度不满足要求.为了继续使用忆阻器阵列,提出一种基于重编程忆阻单元数量约束的闭环重映射算法.首先根据忆阻器阵列的老化分布得出行偏差矩阵;然后以行偏差矩阵中的最小值为起始点开始映射,直至重映射关系形成闭环;通过在映射过程中设置行偏差约束,使得重映射后的行偏差总和尽可能小,达到提高计算精度的目的;通过对重编程单元数量进行约束,尽可能减少需要重新编程的忆阻单元数量,减轻重编程造成的忆阻器阵列老化.在Pytorch上采用MINST数据集进行仿真测试的实验结果表明,所提算法不仅能够有效地提高忆阻器阵列的计算精度,而且与国际上同类方法相比,在达到相同计算精度的前提下,最多可以减少75.43%的重编程单元数.     

高速连续数据记录系统的浮动校验组方法

在很多实时应用领域，如遥感遥测、雷达数据采集中都需要高速连续数据记录系统。文章针对基于盘阵列结构的记录系统，提出了一种提高其可靠性的方法浮动校验组方法。与传统容错方法相比，浮动校验组方法通过阵列结构的动态改变，可以显著提高盘阵列记录系统的可靠性。     

基于Fluent的超声波换能器测风阵列模型改进研究

针对垂直相交型阵列易受阴影效应影响的问题,根据时差法原理,采用三组超声波换能器互成120°放置,设计了一种具有三条独立测风路径的改进型测风阵列,并建立了风速计算模型。该阵列保证至少有两组换能器正常工作的前提下,对某一条受到绕流干扰严重的测风路径进行校正。利用SpaceClaim软件建立测风阵列及风洞模型,通过Meshing软件进行网格划分,使用FLUENT软件改变模拟风洞内的风速及雷诺数等指标,仿真两种阵列在不同流态下的性能表现,得到速度分布云图以及测风路径速度曲线图,说明了改进型测风阵列能明显提升测风精度。     

A spectral imaging array biosensor and its application in detection of leukemia cell

A new type of array immunosensor was developed by combining surface plasmon resonance (SPR) and spectral imaging techniques. The system consisted of a monochromator as the wavelength scanning light source, a polarizer, Kretschmann-Raaether attenuated total reflection (ATR) configuration including array sensor chip, and a CCD camera. The images of transmitting light from ATR were recorded versus the wavelength. By averaging gray scales of the pixels in the area of every gold spot from the image series, the complete spectral resonance curve of all sensing spots on the array can be extracted in parallel. The performance of the developed system was evaluated by analyzing interactions of the anti-CD33 monoclonal antibody to its target leukemic cells using 11 cases of human bone marrow specimens. The specimens were also analyzed with flow cytometry method (FCM) for comparison. The initial results measured by the immunosensor array were corresponded with that of FCM, indicating that the developed parallel method might be clinically suitable for immunophenotyping of acute leukemias. The new sensor array system showed the merits of high-throughput, high sensitivity, high specificity, label free and operation convenient. Spots numbers of the array could be increased if suitable technology were adopted for manipulating the micro bio-liquids on the sensor array chip.     

A novel high degree of freedom sparse array with displaced multistage cascade subarrays

The coprime array is a recently developed sparse array that is widely used in direction-of-arrival (DOA) estimation. The degree of freedom (DOF) for existing coprime arrays relies on the virtual array model, which is limited to the array structure. Furthermore, the continuous virtual array aperture is restricted. This paper aims to propose a novel high DOF sparse array with displaced multistage cascade subarrays to further increase the length of the continuous virtual array and improve the DOF. Through cascading and separating the same sparse arrays, the continuous virtual array aperture is improved and the accuracy of the direction-of-arrival estimation is increased. It is proved that the range of the displaced distance for the maximized continuous virtual array can be obtained. The comparison between the array configuration presented in this paper and the existing common arrays are analyzed. The simulation experiments show that, compared with the uniform array, the coprime array, the nested array, and the coprime array with displaced subarrays (CADiS), the sparse array configuration proposed in this paper can significantly increase the aperture of the continuous virtual array, greatly improve the array DOF and the DOA estimation accuracy, and effectively estimate the DOA of multiple sources in underdetermined conditions.     

WindowsNT与NetWare的互连互访

本文主要介绍ＷｉｎｄｏｗｓＮＴ网络和ＮｏｖｅｌＮｅｔＷａｒｅ网络的互连互访,包括ＮＴ访问ＮｅｔＷａｒｅ服务器和ＮｅｔＷａｒｅ用户访问ＮＴ域两个方面。内容涉及ＮｅｔＷａｒｅ网关服务;ＮｅｔＷａｒｅ客户软件;在ＮＴ服务器上安装、配置ＮｅｔＷａｒｅ网关;在ＮＴＳｅｒｖｅｒ上查看ＮｅｔＷａｒｅ服务器;ＮｅｔＷａｒｅ客户访问ＮＴ域;文件和打印服务ＦＰＮＷ的安装、配置等。     

基于Artix-7系列的eMMC阵列控制器设计与实现

为实现Artix-7系列采集平台存储速率的实时切换,提出了一种具有独立配置接口的eMMC阵列控制器。用户利用配置接口不仅可以提高存储控制逻辑的设计效率,还可以实现基于多种工作频率和阵列操控方式的备用方案,相比于其他同类控制器,集成可靠性更高,应用灵活性更强。首先介绍了eMMC 5.1协议中开放终点多块写、预设块数多块读以及数据擦除的操作原理和技术关键点,然后介绍了总体设计和主要模块的控制流程,最后利用自主设计的测试模块在硬件平台上完成速率切换功能测试和读写速率性能测试。测试结果表明,在200 MHz工作频率下控制器可长时间稳定工作,实现了200 MB/s与800 MB/s之间的速率切换,每片eMMC的最高读写速率可达到200 MB/s。     

On the theoretical link between design parameters and performance in cross-flow fans: a numerical and experimental study

Cross-flow fan performance strictly depends on the complex configuration of the non-axisymmetrical flow field within the machine. The flow field, in turn, is deeply influenced by the design parameters of both casing and impeller geometry. In this paper, the relationship between the design parameters of the geometrical configuration and fan performance is discussed in a theoretical perspective, analyzing the features of the corresponding flow fields. These are reconstructed by a numerical study on cross-flow fan operation carried out for a representative set of configurations at different throttling conditions. Time-accurate solutions for a two-dimensional viscous and incompressible model of the fan using a sliding mesh technique are calculated with a commercial CFD code. The numerical results are validated with experimental data obtained from tests on performance and from local measurements of the flow field.     

基于析取图的非周期可重构流水作业的建模与优化

针对可重构制造这一新的制造模式,研究一类非周期可重构流水作业的建模与优化问题．非周期可重构流水作业由析取图模型表示,基于析取图模型,选取完成所有作业的最大时间Cmax作为优化的目标,建立相应的混合整数规划求解置换、一般和无缓冲区3种情况下非周期可重构流水作业的最优配置和最优调度．     

Improved concentration and separation of particles in a 3D dielectrophoretic chip integrating focusing, aligning and trapping

This article presents a dielectrophoresis (DEP)-based microfluidic device with the three-dimensional (3D) microelectrode configuration for concentrating and separating particles in a continuous throughflow. The 3D electrode structure, where microelectrode array are patterned on both the top and bottom surfaces of the microchannel, is composed of three units: focusing, aligning and trapping. As particles flowing through the microfluidic channel, they are firstly focused and aligned by the funnel-shaped and parallel electrode array, respectively, before being captured at the trapping unit due to negative DEP force. For a mixture of two particle populations of different sizes or dielectric properties, with a careful selection of suspending medium and applied field, the population exhibits stronger negative DEP manipulated by the microelectrode array and, therefore, separated from the other population which is easily carried away toward the outlet due to hydrodynamic force. The functionality of the proposed microdevice was verified by concentrating different-sized polystyrene (PS) microparticles and yeast cells dynamically flowing in the microchannel. Moreover, separation based on size and dielectric properties was achieved by sorting PS microparticles, and isolating 5 μm PS particles from yeast cells, respectively. The performance of the proposed micro-concentrator and separator was also studied, including the threshold voltage at which particles begin to be trapped, variation of cell-trapping efficiency with respect to the applied voltage and flow rate, and the efficiency of separation experiments. The proposed microdevice has various advantages, including multi-functionality, improved manipulation efficiency and throughput, easy fabrication and operation, etc., which shows a great potential for biological, chemical and medical applications.     

Review of formulations for structural and mechanical system optimization

Alternative formulations for optimization and simulation of structural and mechanical systems and other related fields are reviewed. The material is divided roughly into two parts. Part 1 focuses on the developments in structural and mechanical systems, including configuration and topology optimization. Here the formulations are classified into three broad categories: (i) the conventional formulation where only the structural design variables are treated as optimization variables, (ii) simultaneous analysis and design (SAND) formulations where design and some of the state variables are treated as optimization variables, and (iii) a displacement-based two-phase approach where the displacements are treated as unknowns in the outer loop and the design variables as the unknowns in the inner loop. Part 2 covers more general formulations that are applicable to diverse fields, such as economics, optimal control, multidisciplinary problems and other engineering disciplines. In these fields, SAND-type formulations have been called mathematical programs with equilibrium constraints (MPEC), and partial differential equations (PDE)-constrained optimization problems. These formulations are viewed as generalizations of the SAND formulations developed in the structural optimization field. Based on the review, it is concluded that the basic ideas of the formulations presented in diverse fields can be integrated to conduct further research and develop alternative formulations and solution procedures for practical engineering applications. The paper lists 187 references on the subject.