光学遥感压缩成像技术

基于Shannon采样定理的传统信息获取系统在高空间、时间和谱分辨率及系统其它性能上存在难以突破的瓶颈,压缩采样理论为提升航天遥感信息获取能力提供了新的思路。基于压缩采样理论的成像技术（压缩成像）将采样、压缩和数据处理3个过程完美的结合在一起,避免了传统遥感成像系统“先采样再压缩”方式带来的传感器和计算资源浪费,是未来光学遥感极具潜力的成像方式。文章在简要介绍压缩采样基本理论的基础上,总结和分析了国际上目前提出的光学压缩成像系统原型,设计开展了3组压缩成像物理实验,特别结合航天遥感需求设计了推扫式压缩成像方案,实验结果验证了压缩采样的基本原理,并为未来光学遥感压缩成像系统的设计提供了借鉴。     

星载SAR成像处理系统中多线程矩阵转置的设计和实现

矩阵转置被广泛地应用在星载合成孔径雷达的地面成像处理中,由于对原始回波数据快速成像的需求,根据现有共享内存系统的特点,提出利用多线程矩阵转置的并行处理方法。在Linux环境下,利用基于Pthread的线程调度方法和封装类,在研究和分析矩阵转置的几种处理方法的基础上,设计和实现多线程矩阵转置。通过在HP Proliant DL580平台上的实验证明,多线程矩阵转置与传统的矩阵转置相比,能充分利用处理器的资源,具有良好的并行扩展性,有利于进行快速成像处理。     

二自由度平面并联机床的柔度与变形分析

文中对一种新型平面并联机床的速度,力与变形关系进行分析,得到其速度、力雅可比矩阵进而得到其柔度矩阵,提出该并联机床柔度评价指标,并给出了该评价指标在工作空间内的分布情况,为该并联机床的设计与应用提供了有益参考。     

糠酮树脂炭及其炭/炭复合材料微观结构与性能研究

通过浸渍/炭化(PIC)工艺制备了糠酮树脂炭块及糠酮树脂基炭/炭复合材料,对其密度分布、力学和热物理性能及微观结构高温演变过程进行了分析研究。结果表明,制备的糠酮树脂炭块及其炭/炭复合材料密度分布较为均匀,其压缩强度分别为33.9、99.4 MPa,室温~1 000℃时平均线膨胀系数分别为3.91×10-6、1.69×10-6K-1;当热处理温度达到2 100℃左右时,Lc值开始大幅度增长,标志着无定形碳向石墨晶体结构转变。在整个热处理过程中,炭/炭材料石墨微晶尺寸增长幅度比纯树脂炭的大,树脂炭块的显微结构为一种高孔隙度的炭结构,但在炭/炭复合材料中树脂炭与纤维之间界面结合良好。     

面向资源约束航天器控制系统的故障检测研究

针对航天器控制系统的闭环特性和星载计算机存储空间和计算能力等资源受限的特点,研究基于互质分解技术和Youla参数化方法的故障检测方法.考虑航天器控制系统动力学方程和运动学方程,建立线性化系统模型,并给出状态空间表达形式;以状态观测器为基础,利用互质分解技术和Youla参数化方法分别研究控制信号和控制误差与残差的关系,进而给出只与控制信号和控制误差相关的残差设计方法;建立卫星闭环姿态控制系统仿真平台对算法进行仿真验证.仿真结果表明:由于在故障检测过程中避免了观测器的并行运行,因此所提方法在保证故障检测性能的前提下减少了计算量.     

Bush有限单元原理及其在航天器结构建模中的应用

利用大量的数值计算确定了MSC.NASTRAN软件中广义弹簧单元（Bush）刚度阵的解析表达式,通过对比Bush元与剪切梁元的刚度阵,推导了Bush元参数与剪切梁刚度参数的对应关系,并利用展开状态的太阳翼结构和大型桁架结构的静力和动力分析验证了推导结果。该研究揭示了Bush有限单元的原理和所有输入参数的物理含义,有助于Bush元在结构有限元分析中的准确使用。     

计算大型实对称特征问题的 Lanczos-QR 算法

为了计算大型实对称特征值问题Ｋｘ＝λＭｘ的少数低阶特征值对,本文给出Ｌａｎｃｚｏｓ－ＱＲ迭代方法。首先,给定初始迭代向量ｖ１,作ｍ步Ｌａｎｃｚｏｓ分解：ＫＶｍ＝ＭＶｍＴｍ＋ｈｍｅｍＴ。取Ｔｍ的ｄ个最大特征值为移位量,对Ｔｍ进行ｄ步带原点位移的ＱＲ分解。然后,修改初始迭代向量ｖ１。迭代地重新开始这一过程,迫使初始迭代向量ｖ１进入需求的特征子空间,从而使残量‖Ｋｘ－θＭｘ‖→０。数值例子表明,该方法收敛性强,且稳定、有效。     

具有多个执行机构Lurie系统的时滞相关稳定性

Ｌｕｒｉｅ时滞控制系统是工程实践中经常遇到的一类重要的非线性控制系统,本文根据微分方程理论,提出了不确定性具有多个执行机构的Ｌｕｒｉｅ时滞控制系统时滞相关稳定的充分性判据,并给出系统衰减速度的估计方法。该方法仅需假定时滞有界,因而其结论适扩用于变时滞的系统。文末给出应用本文结果的两个应用实例,并与相关成果进行了比较,计算结果表明了本文方法的有效性。     

Investigation on surface roughness,residual stress and fatigue property of milling in-situ TiB2/7050Al metal matrix composites

For higher efficiency and precision manufacturing, more and more attentions are focused on the surface roughness and residual stress of machined parts to obtain a good fatigue life. At present, the in-situ TiB₂/7050Al metal matrix composites are widely researched due to its attractive properties such as low density, good wear resistance and improved strength. It is of great significance to investigate the machined surface roughness, residual stress and fatigue life for higher efficiency and precision manufacturing of this new kind material. In this study, the surface roughness including two-dimensional and three-dimensional roughness, residual stress and fatigue life of milling in-situ TiB₂/7050Al metal matrix composites were analyzed. It was found from comparative investigation that the three-dimensional surface roughness would be more appropriate to represent the machined surface profile of milling particle reinforced metal matrix composites. The cutting temperature played a great role on the residual stress. However, the effect of increasing cutting force could slow down the transformation from compressive stress to tensile stress under 270 °C. An exponential relationship between three-dimensional roughness and fatigue life was established and the main fracture mechanism was brittle fracture with observation of obvious shellfish veins, river pattern veins and wave shaped veins in fracture surface.     

FAST FEATURE RANKING AND ITS APPLICATION TO FACE RECOGNITION

A fast feature ranking algorithm for classification in the presence of high dimensionahty and small sample size is proposed. The basic idea is that the important features force the data points of the same class to maintain their intrinsic neighbor relations, whereas neighboring points of different classes are no longer to stick to one an- other. Applying this assumption, an optimization problem weighting each feature is derived. The algorithm does not involve the dense matrix eigen-decomposition which can be computationally expensive in time. Extensive exper- iments are conducted to validate the significance of selected features using the Yale, Extended YaleB and PIE data- sets. The thorough evaluation shows that, using one-nearest neighbor classifier, the recognition rates using 100-- 500 leading features selected by the algorithm distinctively outperform those with features selected by the baseline feature selection algorithms, while using support vector machine features selected by the algorithm show less prominent improvement. Moreover, the experiments demonstrate that the proposed algorithm is particularly effi- cient for multi-class face recognition problem.