Journal of Systems Engineering and Electronics ›› 2011, Vol. 33 ›› Issue (12): 2593-2597.doi: 10.3969/j.issn.1001-506X.2011.12.05

• 电子技术 • 上一篇    下一篇

适用于模型失配时的改进IMM算法

陈映, 程臻, 文树梁   

  1. 北京无线电测量研究所, 北京 100854
  • 出版日期:2011-12-19 发布日期:2010-01-03

Modified IMM algorithm for unmatched dynamic models

CHEN Ying, CHENG Zhen, WEN Shu-liang   

  1. Beijing Radio Measurement Institute, Beijing 100854, China
  • Online:2011-12-19 Published:2010-01-03

摘要:

机动目标难以跟踪的主要原因是无法找到一个准确的模型来描述目标的运动,即此时目标运动模型是失配的。现今交互式多模型(interacting multiple-model, IMM)算法是一种常用的用于机动目标的跟踪算法。推导分析了现有的典型IMM滤波算法在跟踪机动目标时存在的不足,提出了一种更适用于运动模型失配情况下机动目标跟踪的改进IMM算法。该算法对在跟踪机动目标时滤波器的新息序列的均值特性进行推导分析,改进了IMM算法中模型概率的计算方法,提高了模型概率计算的准确性,从而提高对机动目标的跟踪精度。建立了典型的机动目标跟踪场景,将改进后的IMM算法和原有的典型IMM算法的跟踪性能进行了对比研究,并对模型转换概率的准确性进行了分析,仿真结果验证该改进算法的有效性。

Abstract:

The biggest difficulty in maneuvering target tracking is that the dynamic model of targets can not be found accurately, which means in most situations the dynamic models are unmatched. Interacting multiplemodel (IMM) is the most popular algorithm for maneuvering target tracking nowadays. The paper analyzes the shortcomings of the current typical IMM algorithm for tracking maneuvering targets, and then puts forward a modified IMM algorithm which has a better performance for tracking maneuvering targets. Based on the analysis of innovation expectation, this modified IMM algorithm calculates the model probability in a revised way, which improves the precision for tracking maneuvering targets obviously. Based on a typical simulation scenario for tracking maneuvering targets, the tracking precisions of the modified IMM and current typical IMM are compared, and accuracies of model probabilities calculations for both algorithms are also studied. The simulation results validate the modified IMM algorithm.