交互式多模型算法的模型集设计  

Model set design for interacting multiple model algorithm

作　　者：刘建书[1] 李人厚[1] 张贞耀[1] 刘云龙[1] 

机构地区：[1]西安交通大学系统工程研究所,西安710049

出　　处：《控制与决策》2007年第3期326-328,332,共4页Control and Decision

基　　金：国家"211"行动计划项目

摘　　要：针对交互式多模型(IMM)算法中模型集的设计问题,提出一种模型集设计方法.在已知真实模式概率分布函数的情况下,根据要求的模型数目,等概率划分模式空间,在每个等概率模式空间上,利用方差的含义获得最大范围地覆盖模式空间的模型集,并给出一个例子说明模型集的设计方法.应用Monte-Carlo仿真,将所设计的模型集用于IMM算法进行目标跟踪.仿真结果显示跟踪精度高,表明了该模型集设计方法的有效性.Focusing on the design of the model set for the interacting multiple model（IMM） algorithm, a model set design method is proposed. On the basis that the cumulative distribution function of the true mode is given, the mode space is partitioned into equally probable regions according to the required number of models. In each equally probable region, a model set that covers the maximum mode space is produced by using the variance implication. In addition, an illustration is provided for the design method of the model set. Through Monte-Carlo simulation, the model set designed is used in the IMM algorithm to track the target. Simulation results show that the track accuracy is higher and the proposed model set design method is valid.

关 键 词：交互式多模型算法 模型集设计 目标跟踪 

分 类 号：TP273.2[自动化与计算机技术—检测技术与自动化装置]