基于聚类的ACM模型门限阈值的自适应算法  

Adaptive Threshold Algorithm of ACM Model based on Clustering Fusion

作　　者：王冬旭[1] 姚佩阳[1] 蔡彬彬[2] 郭威武[1] 

机构地区：[1]空军工程大学电讯工程学院,西安710077 [2]长春理工大学,长春130022

出　　处：《火力与指挥控制》2010年第10期40-42,共3页Fire Control & Command Control

基　　金：军队科研基金资助项目(KJ06104)

摘　　要：准确地设置门限阈值是有效提高ACM模型跟踪性能的关键。基于无监督聚类理论,提出一种根据目标机动情况自动确定聚类类别数,通过对实时数据聚类区分出目标做匀速或匀加速运动状态时刻的数据集。再通过加权求解类内样本点的标准差来确定ACM模型的门限阈值。该方法能够根据实时数据自适应的确定门限阈值,计算简便。仿真实验证明该方法能自动确定类别数且分类准确,对噪声标准差估计准确。Accurately set threshold is the key to effectively improve tracking performance of ACM model.In this paper unsupervised clustering method is applied to measurement of maneuvering target,then fuse the measurements in the same cluster,and the number of cluster is determined by target＇s orbit.The threshold of ACM model is set by weighted variance.Practice data show that partitioned clusters is accurate and the algorithm is very effective way of setting threshold of ACM model.

关 键 词：无监督聚类 ACM模型 自适应门限阈值 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]