直接还原在当代冶金变革中的发展

简述了当代的直接还原现状与方法，论述了直接还原是发展短流程的重要基础，直接还原也是熔融还原法的重要组成，同时分析了发展直接还原与熔融还原是当代世界冶金变革中的重要方向之一，也是中国面向２１世纪的重大任务。     

基于AP2DR2指挥信息系统的安全防护体系研究

本文首先介绍了指挥信息系统安全防护的基本概念和特点,然后从网络安全技术和网络安全管理等方面分析了如何实现指挥信息系统完全,提出了一种基于AP2DR2模型的指挥信息系统安全防护体系,论述了AP2DR2安全模型是一个多层次、全方位的动态防御体系,指挥信息系统安全防护体系从静态、被动向动态、主动转变,有效地保障了指挥信息系统信息的安全。     

A discussion of the methods for comparing the relative effectiveness of phosphate fertilizers varying in solubility

Because various phosphate (P) fertilizers differ widely in their solubility, it is commonly observed that crop response to P fertilizers varies under the same soil and crop conditions. Furthermore, a major problem encountered in the methods for determining the relative effectiveness (RE) of water-insoluble P fertilizer (e.g., phosphate rock) with respect to water-soluble P fertilizers, e.g., single superphosphate (SSP) and triple superphosphate (TSP), is that their growth response curves are usually nonlinear and often do not share a common maximum yield. In this paper, we review and discuss the advantages and disadvantages of the three most commonly used methods for calculating the RE of phosphate rock with respect to TSP (or SSP). The three methods are vertical comparison, horizontal (substitution rate) comparison, and linear-response comparison.     

基于双迭代聚能量字典学习的数据压缩算法

针对基于稀疏表示（Sparse representation,SR）的数据压缩压缩率低、重构精度低等问题,本文提出一种基于双迭代的聚能量字典学习算法,把高维信号映射到低维特征空间,当低维特征空间保留高维原始信号越多的特征时,高维信号从低维特征空间中恢复出来的精度越高。为了使低维字典保留高维字典更多的主成分,本文提出了一个新的变换,被命名为?变换,能提升高维字典的能量集中性。除此之外,针对高维字典与低维字典的耦合关系,建立了双循环迭代训练,增加字典的能量集中性与字典的表达能力。实验表明,相比于传统算法,本文提出算法字典学习收敛速度提升了3倍以上。此外,该方法可以得到较高的压缩比和更高质量的重构信号。     

基于生成式对抗网络的图像超分辨率重建算法

针对传统图像超分辨率重建算法存在网络训练困难与生成图像存在伪影的问题,提出一种利用生成式对抗网络的超分辨率重建算法。去除生成式对抗网络的批量归一化层降低计算复杂度,将其中的残差块替换为密集残差块构成生成网络,使用VGG19网络作为判别网络的基础框架,以全局平均池化代替全连接层防止过拟合,引入纹理损失函数、感知损失函数、对抗损失函数和内容损失函数构成生成器的总目标函数,利用纹理损失增强局部信息匹配度,采用激活层前的特征信息计算感知损失获取更多细节特征,使用WGAN-GP理论优化网络模型的对抗损失加速收敛,运用内容损失提升图像低频信息的准确性。实验结果表明,该算法重建图像的平均峰值信噪比为27.97 dB,平均结构相似性为0.777,与SRGAN和EDSR等算法相比,其在未延长较多运行时间的情况下,重建结果的纹理细节更清晰且亮度信息更准确,更符合视觉感官评价要求。     

学习小波超分辨率系数的人脸超分算法

目前基于卷积神经网络的超分方法虽然在峰值信噪比和结构相似性评价指标上能取得优异的结果,但是得到的超分图像视觉质量较差,会丢失人脸五官区域的细节信息。针对这一现象,设计了一种新的深度神经网络来预测超分小波系数以获得信息丰富的超分辨率人脸图像,首先利用人脸图像的先验知识手动地给予五官区域更多的关注,然后在网络中引入线性低秩卷积运算,最后利用长距离依赖的思想补充超分图像的细节。实验验证该算法可以在获得较高的峰值信噪比和结构相似性的同时,使超分人脸图像五官区域更加清晰、视觉质量更优。     

Finishing effects of spiral polishing method on micro lapping surface

This study presents a spiral polishing method and a device for micro-finishing purposes. This novel finishing process has wider application than traditional processes. This offers both automation and flexibility in final machining operations for deburring, polishing, and removing recast layers, thereby producing compressive residual stresses even in difficult to reach areas. Applying of this method can obtain a fine polished surface by removing tiny fragments via a micro lapping generated by transmission of an abrasive medium through a screw rod. The effect of the removal of the tiny fragments can be achieved due to the function of micro lapping. The method is not dependent on the size of the work-piece's application area in order to carry out the ultra precise process. The application of this research can be extended to various products of precision ball-bearing lead screw. The proposed method produces products with greater precision and more efficiently than traditional processes, in terms of processing precisions and the surface quality of products. These parameters used in achieving maximum material removal rate (MRR) and the lowest surface roughness (SR) are abrasive particle size, abrasive concentration, gap, revolution speed and machining time.     

结合稀疏编码模型的多帧图像超分辨率重建

传统序列超分辨率方法对低分辨率视频序列的要求较高,一旦序列中没有包含足够的信息,会造成重建高分辨率图像质量的下降。为此,提出一种结合稀疏编码模型的序列超分辨率算法。利用概率运动场从低分辨率序列中重建一幅高分辨率图像,根据自适应阈值确定重建有效和无效区域,使用稀疏编码模型对无效区域进行补全重建。实验结果表明,该算法可以采用序列自身的信息和稀疏字典中的信息来重建高分辨率图像,在序列信息有破缺时,与仅利用序列自身信息或仅利用单幅图像的算法相比,具有更好的鲁棒性和广泛的适用性。     

软件无线电中前端数字下变频的研究

本文在讨论软件无线电的基本原理、结构、特点以及关键技术的基础上,研究了软件无线电数字前端中的数字下变频(DDC)技术。     

Magnetic order and evolution of the electronic state aroundx=0.12 in La2?xBaxCuO4 and La2?xSrxCuO4

Muon spin rotation ( ⁺SR) measurement provides clear evidence of the antiferromagnetic order of Cu moments below 35 K for La_2–x Ba _x CuO₄ and below 15 K for La_2–x Sr _x CuO₄ in the narrow range ofx where the high-T _c superconductivity (SC) is suppressed remarkably. The results suggest that the change of the electronic state coupled with the lattice instability is relevant to the local suppression of SC and freezing of spin fluctuations of the Cu moment.