目标预测中的野值剔除方法研究  

Outlier-discrimination Algorithm in Target-prediction

作　　者：卢元磊[1] 何佳洲[1] 安瑾[1] 

机构地区：[1]江苏自动化研究所,连云港222006

出　　处：《计算机与数字工程》2013年第5期722-725,共4页Computer & Digital Engineering

摘　　要：目前在目标跟踪与预测中野直判别方法单一,普遍采用3σ准则。当目标发生机动时,现有的野值剔除准则存在野值误剔除率过高的问题。论文针对这个问题开展研究,首先,分析了几种常用准则的特点;然后提出了不确定观测点的概念,并设计了能够适应目标机动的野值剔除算法。仿真结果表明:论文算法有效的降低了野值处理中的误剔除率,提高了目标跟踪预测的精度。For current target tracking,target prediction is unitary in the method of outlier discrimination,adopting commonly used 3σ law which is difficult to discriminate target maneuver and measure outliers.It is difficult to handle outliers,when the target maneuvers.Focusing on this problem,this algorithm puts forward the concept of uncertain observation and an algorithm is preseneed based on feedback controller.The simulation test shows that the proposed algorithm can effectively distinguish and eliminate outliers among the observed data.It decreases the false outlier elimination rate and improves the performance of the prediction.

关 键 词：野直剔除 目标机动 目标预测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]