基于遗传算法的鲁棒H_∞控制系统设计

提出了一种基于遗传算法的H∞ 控制系统设计方法 ,把控制系统的多个设计目标转化成不等式约束 ,将H∞ 控制器设计中加权阵的选取表示成一个多目标优化问题 ,通过遗传算法优化加权阵参数 ,进而找到同时满足给定的频域和时域性能指标要求的H∞ 控制器。以UH - 6 0A直升机为对象 ,采用此方法设计出了直升机的鲁棒飞行控制器 ,使得所设计的直升机飞行控制系统不仅具有鲁棒性 ,而且满足所有给定的性能指标要求。     

On the polymorphism of cyanogenesis in Lotus corniculatus L.

Most populations of the legume Lotus corniculatus are polymorphic for the characters of leaf cyanogenesis and petal cyanogenesis. The distribution of these phenotypes was investigated in a section of chalk quarry where the vegetation ranged from pioneer species growing on almost bare chalk through to a mature grassland assemblage. The vegetation surrounding randomly chosen L. corniculatus individuals was recorded and this information was used to compare the environments in which the different phenotypes were growing. Leaf cyanogenic plants were found to occur at higher frequencies in grassland areas than in the more open conditions typical of earlier stages in the succession at the quarry. Factors that may be responsible for the distribution of the phenotypes are discussed.     

列车占线问题的无交叉遗传算法

本文就列车占线问题,针对一般遗传算法的“早熟收敛”,提出了一种无交叉遗传算法,并通过实例仿真验证了算法的有效性。     

波分复用星形网的波长重用扩容

针对有限的数据信道波长对波分复用星形单跳网容量限制,提出了一种波长重用和多耦合器内连相结合的组网扩容方法.该方法不但可将网络所支持的节点数(网络容量)扩大一倍;而且网络的所有数据信道波长都得到重用,网络的吞吐量增加一倍.在网络节点数不变的条件下,通过波长重用,可大大减少网络中通信节点的排队时延,缓和各通信节点对数据信道波长使用权的竞争矛盾,有效地改善网络性能.此外,该方法结构简单,易于实施.     

增强型模糊形态学联想记忆的改进

为了克服增强型模糊形态学联想记忆（EFMAM）的网络规模随训练样本数目的增加而急剧增大的缺点，使其硬件实现更为容易，对原始EFMAM的网络体系结构进行了约简．在相似度空间中，定义一个基于概率（“软”）相似度分配的准则以评价约简矩阵A，再利用遗传算法搜索其最佳解．实验表明，约简后的EFMAM节约了计算时间和存储空间，易于硬件实现，同时其识别性能相对于原始EFMAM并未明显降低．     

Molecular phylogenetic status of Siberian roe deer （Capreolus pygargus） based on mitochondrial cytochrome b from Jeju Island in Korea

In order to elucidate the genetic diversity and taxonomic status of Siberian roe deer （Capreolus pygar- gus） from the Korean peninsula, partial sequences of the mitochondrial cytochrome b gene were analyzed. We col- lected 57 roe deer samples from Mainland and Jeju Island between July 2004 and June 2009. A total of 50 sequences （1,070 bp） were obtained, and seven haplotypes were defined. Five haplotypes were found from Mainland, and two were discovered from Jeju Island. The genetic diver- sity was lower in the Jeju Island population than in the Mainland population. In addition, both the consensus neighbor-joining tree and maximum parsimony tree showed an identical topology between the Mainland and Jeju Island clades. Based on the results, two subspecies of the Korean roe deer were confirmed. Siberian roe deer in Jeju Island showed a distinct difference in the cytochrome b gene compared with other Siberian roe deer, being aunique native species inhabiting only Jeju Island in Korea. Because Siberian roe deer in Jeju Island can be classified at subspecies level, the new subspecies in Jeju was named Capreolus pygargus jejuensis in this study. For this new subspecies, reasonable conservation and management effort is needed in the future.     

考虑提前／拖期成本的计划提前期确定

研究了随机提前期情况下物料计划提前期的确定问题。以提前／拖期成本最小为目标，通过构造遗传算法的适应值函数寻求一组最优的计划提前期，用参数设置模拟应用中的实际情况，使物料的提前／拖期成本最小。最后通过算例证明了算法的有效性。     

中国鹿科动物遗传资源的现状

本文分析了我国鹿科动物的遗传资源现状及其导致原因，并阐明遗传资源优势，提出抢救性保护方法和合理利用途径。     

中国区域科技资源配置效率及影响因素分析

以RD人员全时当量等6个指标为科技资源投入,以专利申请受理量等5个指标为科技资源产出,应用熵权法计算各全国各地区2008-2015年的科技投入、产出的权重,应用指标体系法来构建科技资源配置效率模型,计算各地区科技资源配置效率及时空演进特点。应用ANN-GA方法,计算出科技资源配置效率与科技资源投入产出组合最优解,并与北京市进行比较,指导提升科技资源配置效率。     

Internet中的多播路由选择算法

Ｉｎｔｅｒｎｅｔ 中多播路由选择问题可归化为最小 Ｓｔｅｉｎｅｒ 树问题，而该问题是一个 Ｎ Ｐ 难题，当前只存在若干种启发式算法。基于遗传算法和最小树算法，提出并实现了一种新的自适应路由选择算法，经编程运行发现，此算法具有良好的实际应用前景。