一种基于聚类分析的数值积分新算法  

A New Numerical Quadrature Algorithm Based on Cluster Analysis

作　　者：王景亮[1] 陈静[1] 罗振东[2] 

机构地区：[1]中国农业大学理学院,北京100083 [2]华北电力大学数理学院,北京102206

出　　处：《世界科技研究与发展》2012年第1期89-92,共4页World Sci-Tech R&D

基　　金：国家自然科学基金(10871022;11061009);中央高校基本业务费专项资金(2009-2-05)资助项目

摘　　要：通过聚类分析找出一般数值求积公式的被积函数出现数值信息波动较大的区域,自适应地采用相应的高精度公式,构造出高精度的自适应数值求积公式。数值试验结果表明,这种新数值求积算法较一般数值求积公式,能显著地减少数据存储量和计算量,提高数值求积公式的稳定性和数值积分的精度。The fields of larger fluctuation in numerical information of integrated function are found by clustering. The high accuracy numerical integral formals are adopted elf-adaptively in the process of solving numerical integral. A new numerical quadrature algorithm is presented which is based on cluster analysis. It is shown by numerical examples compared with the traditional algorithms that the 6omputation time and data storage can be greatly reduced by the new algorithm, with the stability and the accuracy of numerical quadrature algorithms improving.

关 键 词：聚类分析 数值积分 预处理 自适应 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]