基于扩展分形和CFAR特征融合的SAR图像目标识别

研究了多信息融合技术在SAR图像目标识别中的应用。将扩展分形特征(Extended Fractal)与双参数恒虚警特征(Double Parameter CFAR)形成的多信息进行融合处理。运用Dempster-Shafer证据理论，在决策层对SAR图像中的像素进行识别分类。实验结果表明通过融合对像素分类的准确性明显好于单特征的检测结果，减少了虚警概率，提高了系统的识别能力。     

可重构制造系统重组过程中设备选择模型的研究

根据生产任务选择加工设备进行制造资源重组是实现可重构制造系统的关键问题之一，由于设备的选择涉及到多种因素，既有定量指标，又有定性指标，传统的依靠人工经验的方法显得力不从心。本文首先结合实际情况，提出了一套设备选择评价体系，通过对模糊判断矩阵采用最小对数二乘法确定各评价因素的权重系数，针对定性指标和定量指标采用不同的方法确定其性能指标值，通过模糊积分对评判指标进行综合评判，最后进行了实例研究。所提出的方法有效地简化了决策过程，为可重构制造系统设备选择提供了一套行之有效的方法。     

与HBV X蛋白相互作用的细胞蛋白的筛选及其鉴定

参考GenBank核苷酸序列库中的HBV的核苷酸序列设计合成了一对对应于HBx基因的引物，从原发性肝癌(hepatocellular carcinoma，HCC)患者的血清中提取HBV基因组DNA作为模板，扩增HBx编码区并测定了核苷酸序列．将该编码区克隆到酵母双杂交系统的诱饵蛋白表达载体pGBKT7中，转入酵母细胞AH109，进行表型鉴定．然后通过Mating实验从已制备好用于酵母双杂交系统的肝cDNA表达文库中筛选与HBx相互作用的细胞蛋白，并用体外免疫共沉淀实验进一步验证．研究表明，分离得到的与HBx基因编码的蛋白相互作用的4种新的细胞蛋白，分别是醛缩酶B、C8α亚基、一种丝氨酸蛋白酶Hepsin和一种未知蛋白．     

Geometrical Molecular Structure Design Concept in Approach to Homo- and Heterometallic Precursors of Advanced Materials in Sol–Gel Technology

The article describes the principles of the Single Source Precursor approach to inorganic materials and introduces the Geometrical Molecular Structure Design Concept (MSDC) based on the choice of a proper molecular structure type for the desired precursor and completing it with ligands providing both the necessary number of donor atoms and the sterical protection of the chosen core. Application of MSDC is illustrated with examples taken from development of new approaches in the synthesis of oxide and sulfide catalysts and ferroelectric oxide materials.     

Self-Interactions of Strands and Sheets

Physical strands or sheets that can be modelled as curves or surfaces embedded in three dimensions are ubiquitous in nature, and are of fundamental importance in mathematics, physics, biology, and engineering. Often the physical interpretation dictates that self-avoidance should be enforced in the continuum model, i.e., finite energy configurations should not self-intersect. Current continuum models with self-avoidance frequently employ pairwise repulsive potentials, which are of necessity singular. Moreover the potentials do not have an intrinsic length scale appropriate for modelling the finite thickness of the physical systems. Here we develop a framework for modelling self-avoiding strands and sheets which avoids singularities, and which provides a way to introduce a thickness length scale. In our approach pairwise interaction potentials are replaced by many-body potentials involving three or more points, and the radii of certain associated circles or spheres. Self-interaction energies based on these many-body potentials can be used to describe the statistical mechanics of self-interacting strands and sheets of finite thickness.     

Alkenylidenes in Organic Synthesis

Alkenylidenes R₂C?C: (= alkylidene carbenes) undergo regio- and stereoselective intramolecular C? H insertion reactions that are excellently suited for the synthesis of cyclopentenes. The 1, 2-shifts occurring with R?H and R?Ar are useful for the preparation of alkynes. Alkenylidenes are efficiently generated from carbonyl compounds by diazomethylation, from vinyl halides by α-elimination, from alkynyliodonium salts by addition of nucleophiles, and from alkynes by retro-1, 2-shifts. Specific applications of the various methods, particularly in the synthesis of natural products, are discussed.     

基于几何不变量的蛋白质三维分子结构比对

为了解决蛋白质三维结构比对需要处理大量的旋转、平移变换，直接用动态规划将变得十分繁琐这一问题，在保留蛋白质空间结构属性特征的基础上，对蛋白质三维数据进行了预先的处理．通过计算蛋白质结构在旋转和平移下的几何不变量，将蛋白质的三维结构坐标变换为具有旋转、平移不变性的一维序列．进一步给出了“距离”以及“相似得分”的定义．在此基础上采用动态规划方法给出了新的蛋白质结构比对算法．对专家分类的蛋白质结构数据库进行测试，结果显示准确、快速．     

现代优化计算方法在蛋白质结构预测中的应用

现代优化计算方法在蛋白质结构预测中占有重要地位.简要地介绍了模拟退火算法,遗传算法,人工神经网络和图论算法在蛋白质结构预测中的应用.对国内外近年来应用这些算法,特别是在蛋白质构象搜索问题中,解决蛋白质结构预测的研究作了回顾,并分析、比较了这几种算法的效果和特点.