基于规划识别的态势估计方法研究  

Research on Situation Estimation Method Based on Planning Recognition

作　　者：李尚昊 方浩[1] LI Shanghao;FANG Hao(School of Automation,Beijing Institute of Technology,Beijing 100081,China)

机构地区：[1]北京理工大学自动化学院,北京100081

出　　处：《无人系统技术》2022年第5期31-38,共8页Unmanned Systems Technology

基　　金：国家自然科学基金(62133002,U1913602)。

摘　　要：针对战场态势估计问题中的目标分群和规划识别问题进行研究。在目标分群问题的研究中,提出一种二次分群算法。首先利用最近邻法进行聚类,然后采用基于模板匹配的方法进行初次分群,考虑到其无法应对空间位置有重叠的情况,结合信代价及初次分群属性沿用模板匹配的思想进行二次分群;在规划识别问题的研究中,提出一种基于贝叶斯网络的规划识别算法,为了获取顶层规划任务并求得概率,构建贝叶斯网络,将当前获得的事件与网络库进行匹配,实现对敌方的规划识别。针对以上两种算法,设计了符合战场情景的仿真环境,通过仿真验证了所提出的二次分群算法相较于初次分群识别准确率提高了26.7%,规划识别算法准确率达98.04%。最后,对所做工作进行了总结并对未来可能的改进方向进行分析。The problem of target clustering and planning identification in battlefield situation estimation is studied.In the research of target clustering,a secondary clustering algorithm is proposed.First,the nearest neighbor method is used for clustering,and then the method based on template matching is used for primary clustering.Considering that it can not cope with the situation of overlapping spatial positions,the idea of template matching is used for secondary clustering in combination with communication costs and primary clustering attributes.In the research of planning identification,a planning identification algorithm based on Bayesian network is proposed.In order to obtain the top-level planning tasks and obtain the probability,a Bayesian network is constructed to match the currently obtained events with the network library to realize planning identification of the enemy.For the above two algorithms,a simulation environment suitable for the battlefield situation is designed.It is verified by simulation that the accuracy of the proposed secondary clustering algorithm is 26.7%higher than that of the primary clustering recognition,and the accuracy rate of the planning recognition algorithm is 98.04%,and the effectiveness of the algorithm is verified by simulation.Finally,the work of this paper is summarized and the possible improvement direction in the future is analyzed.

关 键 词：态势估计 目标分群 规划识别 聚类 贝叶斯网络 

分 类 号：E91[军事]