基于模糊推理的自适应交互多模型目标跟踪算法  

Adaptive IMM Tracking Algorithm Based on Fuzzy Inference

作　　者：朱志宇[1] 

机构地区：[1]江苏科技大学电子信息学院,江苏镇江212003

出　　处：《弹箭与制导学报》2008年第1期29-32,36,共5页Journal of Projectiles,Rockets,Missiles and Guidance

基　　金：国防预先研究基金资助

摘　　要：提出了一种模糊自适应IMM算法(FAIMM),通过模糊逻辑,根据加速度估计值,自适应地摒弃概率较小的模型,仅选取整个模型集合中最能反映目标"当前"机动的一个模型子集进行运算,从而减少了模型数目;同时采用模糊方法计算模型概率,从而降低了算法的计算量;进一步,通过二级模糊推理,根据模型参考加速度ui的大小自适应地选择适当的最大机动加速度amax和a-max,使系统具有一定的方差调整能力,从而提高了跟踪精度。仿真结果表明,较之于标准IMM算法,FAIMM算法在机动目标跟踪精度、跟踪的平稳性以及收敛速度等方面都有所改善。A fuzzy adaptive IMM（FAIMM） algorithm was proposed to adaptively design model set of IMM in this paper, which acquires posterior probability of model through fuzzy inference based on estimation of acceleration, in this way the models having small probability were adaptively eliminated from the whole model set, only one model subset who could reflect ＂current＂ maneuver of target best was reserved so that model number was decreased; meanwhile the probability of mode was acquired through fuzzy method, resulting in smaller computation amount. Second level fuzzy inference was applied to adaptively choose maximal maneuvering acceleration amax and a-max according to reference acceleration ui of model, so that tracking system had certain ability to adjust covariance to improve tracking precision. Simulation results indicate that FAIMM has higher precision, better stability and faster convergence speed than IMM algorithms in application of maneuver target tracking.

关 键 词：IMM算法 模糊 “当前”统计模型 机动目标跟踪 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] TP391[自动化与计算机技术—控制科学与工程]