摘要
在进行海上作战态势分析时,通常需要剔除目标海域内对态势分析影响较小的非作战目标,只保留参考价值较高、作战相关的目标航迹数据。现有的行为规律挖掘方法大多是基于聚类的思想,作用于非作战目标清洗问题时工作步骤较为复杂、效果较差。结合态势分析需求,基于相似重复记录检测的思想,通过定义多维度记录匹配相似度(multi-dimension record similarity,MDRS),提出了一种海上非作战目标实时清洗方法。通过对多维航迹数据的相似重复检测,实现对非作战目标的实时清洗。在仿真军事场景上进行实验分析,结果表明所提方法能够实时、有效地检测出海上非作战目标。
Objectives: It is usually necessary to eliminate non-combat targets for conducting marine combat situation analysis. Because non-combat targets have little impacts on situational analysis and the retained combat-correlated targets have high reference value. Most of the existing behavior mining methods, which are based on the idea of clustering, are complicated and ineffective in cleaning of non-combat targets.Methods: Therefore, this paper defines multi-dimension record similarity(MDRS) and proposes a real-time cleaning method for marine non-combat targets(MNCT-RTCM). The proposed method realizes the real-time cleaning of non-combat targets by similar duplicate record detection of multi-dimensional track data. Results: The experiments are carried out on simulated military scenarios, and the results are evaluated and analyzed by calculating the recall rate and the precision rate. Conclusions: The results show that the MNCT-RTCM method can effectively detect non-combat targets and achieve real-time cleaning of noncombat targets in the marine combat environment.
作者
林雪原
李雪腾
潘新龙
李敏波
陈祥光
LIN Xueyuan;LI Xueteng;PAN Xinlong;LI Minbo;CHEN Xiangguang(Department of Electrical and Electronic Engineering,Yantai Nanshan University,Yantai 265713,China;91827 Troops,Weihai 264200,China;Institute of Information Fusion,Naval Aviation University,Yantai 264001,China;Software School,Fudan University,Shanghai 200433,China)
出处
《武汉大学学报（信息科学版）》
EI
CAS
CSCD
北大核心
2021年第9期1378-1385,共8页
Geomatics and Information Science of Wuhan University
基金
国家自然科学基金(91538201,61671157)
烟台市“双百计划”人才项目(YT201803)。
关键词
态势分析
非作战目标
实时清洗
相似重复记录
situation analysis
non-combat targets
real-time cleaning
similar duplicate record
