基于模糊ART划分的目标分群算法  

A Fuzzy ART Based Target Clustering Algorithm

作　　者：樊振华[1] 师本慧[1] 陈金勇[1] 段同乐[1] 王长力[2] FAN Zhen-hua SHI Ben-hui CHEN Jm-yong DUAN Tong-le WANG Chang-li(The 54th Research Institute of CETC, Shifiazhuang Hebei 050081, China Unit 91635, PLA, Beijing 102249, China)

机构地区：[1]中国电子科技集团公司第五十四研究所 [2]中国人民解放军91635部队

出　　处：《无线电工程》2017年第9期27-31,共5页Radio Engineering

基　　金：海洋公益性科研专项基金资助项目(201505002)

摘　　要：针对目标分群中存在的分类数未知和噪声干扰问题,提出一种基于模糊ART划分的目标分群算法。通过目标识别属性划分,约减分群目标数规模,降低计算量;通过划分数据预处理消除尺度差异,在此基础上采用基于模糊ART的目标空间划分,经类选择、匹配度检验和类学习等步骤实现对目标的增量式动态分群。试验结果表明,该算法对复杂环境下未知分类数的多目标编队分群具有良好的有效性、稳健性和实时性。Target clustering is confronted with problems of unknown category number and noise interference. To solve these problems,a fuzzy ART based target clustering algorithm is proposed.In this algorithm,the clustering target number is reduced through attribute partition to lessen the computational burden.Through data preprocessing,the scale difference is eliminated.On this basis,target space partition based on fuzzy ART is adopted, incremental dynamic clustering of targets is achieved through category choice, resonance and learning. Experimental results show the effectiveness, stability and real-time performance of the proposed algorithm in multi- formation clustering with unknown category number in complicated environment.

关 键 词：目标分群 空间划分 模糊ART 属性划分 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]