论文部分内容阅读
随着现代战争节奏的加快,以及传感器所产生战场监视数据量的剧增,指挥人员面临着越来越大的认知压力.目标分群作为一种重要的高级数据融合技术,能够减轻指挥员的认知负担,但目前的目标分群算法都需要人为地给出一些参数,且分群的依据较为单一,不能满足联合作战指挥的需要.为解决多属性目标分群问题,首先确定了目标分群问题的描述方式,其次以多目标属性为基础,通过计算相似度和网络最佳分类判定函数,提出了基于层次聚类的非监督目标分群算法,最后给出了算法的实现描述和例子.
As the rhythm of modern warfare accelerates and the amount of battlefield surveillance data generated by sensors increases, commanders face increasing cognitive pressure.As an important advanced data fusion technology, target grouping can reduce the risk of commanders However, the current target-based clustering algorithms all need to give some parameters artificially, and the basis of grouping is relatively simple and can not meet the requirements of joint operations command. To solve the multi-attribute target grouping problem, firstly, the description of target grouping problem Secondly, based on multi-objective attributes, this paper proposes an unsupervised target clustering algorithm based on hierarchical clustering by calculating the similarity and the best network classification and decision function. Finally, the implementation description and examples of the algorithm are given.