A Battlefield Target Grouping Method Based on M-CFSFDP Algorithm

基于M-CFSFDP算法的战场目标分群方法

¹ School of Automation, Northwestern Polytechnical University, Xi’an 710072, China
² AVIC Xi'an Flight Automatic Control Research Institute, Xi'an 710065, China

Received: 25 December 2017

Abstract

Target grouping can divide battlefield targets into battle space groups. In this way, the target grouping reduces the difficulty of situation assessment and increases the efficiency of decision. In order to solve the target grouping, a target grouping method based on Manifold-CFSFDP algorithm is proposed. This method turns target grouping into dataset clustering. After calculating the manifold which measures the similarity of targets, it searches the clustering centers and classifies the other data points by CFSFDP based on manifold. The simulation experiment for artificial and UCI datasets proves that M-CFSFDP is more effective than CFSFDP. The correctness and feasibility of M-CFSFDP are also shown by static and dynamic grouping of battlefield targets.

摘要

目标分群能够将战场目标划分为作战空间群，从而降低态势估计难度，提高决策效率。故针对战场中的目标分群问题，提出了一种基于流形距离（manifold）的密度峰值快速搜索聚类算法（clustering by fast search and find of density peaks，CFSFDP）的目标分群方法。该方法将目标分群转化为数据集聚类问题，通过计算目标间的流形距离来衡量目标间的相似度，然后在流形距离的基础上利用CFSFDP算法搜索聚类中心，指定其余数据点类别。仿真实验以人工数据集和UCI数据集为对象，验证了M-CFSFDP算法聚类效果优于CFSFDP算法；同时将M-CFSFDP应用在战场目标静态与动态分群中，仿真结果表明了该方法的正确性与有效性。