基于改进空间划分的目标分群算法
发布时间：2018-11-04 07:47
【摘要】：针对战场目标分群中存在的类数未知和阈值选取欠缺有效方法的问题,提出一种基于改进空间划分的目标分群算法。首先,通过敌我及作战单位属性划分,约减分群目标数规模,降低计算量;其次,通过对空间距离划分进行改进,能够动态地优选阈值,有效解决类数未知的分群问题。通过引入划分独立性和逆χ~2分布概率区间约束,消除计算冗余并提取出候选阈值,在此基础上选取最大的候选阈值作为最终分群阈值,可以有效滤除过程噪声与观测噪声干扰,提高分群准确率。仿真结果表明,该算法对战场环境下的多目标编队分群具有良好的有效性、稳健性和实时性。
[Abstract]:In order to solve the problem of unknown class number and lack of effective method of threshold selection in battlefield target clustering, a new algorithm based on improved space partition for target clustering is proposed. First, by dividing the attributes of the enemy and the other and the operational units, the size of the number of targets in the cluster is reduced, and the computation is reduced. Secondly, by improving the space distance division, the threshold can be dynamically selected and the clustering problem with unknown number of classes can be effectively solved. By introducing partition independence and inverse 蠂 ~ 2 distribution probability interval constraints, the computational redundancy is eliminated and candidate thresholds are extracted. On this basis, the largest candidate threshold is selected as the final clustering threshold, and the process noise and observation noise interference can be effectively filtered. Improve the accuracy of clustering. Simulation results show that the algorithm is effective, robust and real-time for multi-target formation clustering in battlefield environment.
【作者单位】： 中国电子科技集团公司第五十四研究所;
【基金】：中国博士后科学基金(2015M580217) 河北省博士后科学基金(B2015005003)资助课题
【分类号】：TP301.6;E91
，
本文编号：2309166
[Abstract]:In order to solve the problem of unknown class number and lack of effective method of threshold selection in battlefield target clustering, a new algorithm based on improved space partition for target clustering is proposed. First, by dividing the attributes of the enemy and the other and the operational units, the size of the number of targets in the cluster is reduced, and the computation is reduced. Secondly, by improving the space distance division, the threshold can be dynamically selected and the clustering problem with unknown number of classes can be effectively solved. By introducing partition independence and inverse 蠂 ~ 2 distribution probability interval constraints, the computational redundancy is eliminated and candidate thresholds are extracted. On this basis, the largest candidate threshold is selected as the final clustering threshold, and the process noise and observation noise interference can be effectively filtered. Improve the accuracy of clustering. Simulation results show that the algorithm is effective, robust and real-time for multi-target formation clustering in battlefield environment.
【作者单位】： 中国电子科技集团公司第五十四研究所;
【基金】：中国博士后科学基金(2015M580217) 河北省博士后科学基金(B2015005003)资助课题
【分类号】：TP301.6;E91
，
本文编号：2309166
本文链接：https://www.wllwen.com/kejilunwen/ruanjiangongchenglunwen/2309166.html
