基于最近邻接点的涡点搜索算法  

Vortex point searching algorithm based on nearest neighbors point

作　　者：刘杰[1] 孙秦[1] 

机构地区：[1]西北工业大学航空学院,陕西西安710072

出　　处：《计算机工程与设计》2013年第3期920-924,共5页Computer Engineering and Design

摘　　要：为了实现对优化迭代初始点(涡点)的预处理,保证从各个初始点迭代得到不同的局部最优点,进而通过对比研究获得全局最优点,提出了以基于Pan-距离的最近邻接点搜索为核心、涡点之间必有凸起为判断准则的涡点搜索算法。该算法以Pan-距离作为高维数据点之间的相似度描述参数。基于Pan-距离的高维空间邻接点搜索算法有效地降低了涡点搜索的计算负担。算例结果表明,在抽样点密度足够的情况下,该算法可实现高维空间涡点的快速有效搜索。To realize the preprocessing of the initial points or called vortex points before optimization iterations and to ensure that different local optimums can be achieved, which makes global optimum obtained easily by comparison of these local optimums, a vortex point searching algorithm is proposed, in which Pandistance based on nearest neighbors searching algorithm is adopted. The criterion is that there must be a peak between two vortex points. Pandistance is regard as characteristic parameter of deseri bing the similarity between multidimensional data in the algorithm. The nearest neighbors searching algorithm for highdimen sional space based on Pandistance reduces the calculation burden of vortex point searching effectively. The numerical results show that when the density of scatter points is sufficient, vortex points in highdimensional space is searched quickly and effi ciently by using this algorithm.

关 键 词：Pan-距离 最近邻接点 涡点搜索 优化 全局最优值 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]