摘要
能否快速精确给出高空飞行器落点信息是判断飞行任务是否成功的重要依据。针对在工程应用中使用常规DBSCAN算法对多组落点预测数据进行统计聚类时,全局参数Eps及min Pts组合对聚类结果有较大影响,且需要依靠先验经验进行人工干预,难以量化处理的问题,基于样本集数据特征与核密度估计原理,在实测落点数据集中仿真实现了两种自适应DBSCAN算法,并进行了适用性分析对比,解决了快速落点预报中DBSCAN算法参数选择不合理导致聚类质量恶化的问题。结果表明:能够合理地选择全局参数Eps及min Pts,效率较优,提高了DBSCAN算法在快速落点预报中的准确率与适用性。
It is an important basis to judge the success of flight mission whether the impact point information of high altitude flights can be given quickly and accurately.In engineering applications,when conventional DBSCAN algorithm is used in the clustering of several groups of impact points,the global parameters Eps and minPts has great influence on the clustering result,meanwhile,the choice of them requires prior information for manual intervention,so it is difficult to quantify.In this regard,based on the own characteristics of data sets and kernel density estimation,two kinds of self-adaptive DBSCAN algorithms are implemented in measured data set,the applicability analysis is also carried out and contrasted.The quality of clustering deteriorates caused by unreasonable selection of global parameters in fast impact point prediction is solved.The result shows that the proposed method can select reasonable global parameters Eps and minPts,the efficiency is good,the applicability and accuracy of DBSCAN algorithm are improved.
作者
陶鹤丹
项树林
吴诗帆
TAO Hedan;XIANG Shulin;WU Shifan(No.91550 Troops of PLA,Dalian 116000)
出处
《舰船电子工程》
2023年第2期84-88,165,共6页
Ship Electronic Engineering
关键词
高空飞行器
快速落点预报
聚类
DBSCAN
自适应
high altitude flights
fast impact point prediction
clustering
DBSCAN
self-adaptive
