基于概率相关性的AIMM跟踪算法  

Adaptive interactive multiple model tracking algorithm based on probability relativity

作　　者：左涛[1] 刘进忙[1] 郑富军[1] 

机构地区：[1]空军工程大学导弹学院,陕西三原713800

出　　处：《电光与控制》2007年第5期168-171,共4页Electronics Optics & Control

摘　　要：为解决传统自适应交互式多模型(AIMM)算法计算量普遍过大而不适应于实际工程应用的问题,提出了基于概率相关性的自适应的交互多模型算法(PR-AIMM),该方法在由CA、CV和基于圆周运动的转弯(TR)模型3个基本模型组成的模型集上,利用模型后验概率最近时间相关性自适应地调整马尔可夫转移矩阵的参数,能对空中大部分机动目标进行有效的跟踪,有效地解决在实际工程应用中机动目标跟踪问题。最后运用仿真试验对上述算法进行了合理性和有效性验证,并在实际工程应用中达到了满意的效果。The traditional Adaptive Interactive Multiple Model （AIMM） algorithm has tremendous calculation cost and is not suitable for practical engineering application. To solve the problem, an AIMM tracking algorithm based on Probability Relativity （PR-AIMM） is proposed. Based on the set composed of the three basic models of CV, CA and TR, the algorithm adjusts Markov transition probabilities by using the latest relativity of back probability. It can implement effective tracking for most of maneuvering target, so as to solve most of target tracking problems in actual project. Simulation proves the rationality and validity of the algorithm. The algorithm has been applied in actual project, and achieved satisfying effect.

关 键 词：AIMM算法 概率相关性 马尔可夫转移概率 目标跟踪 

分 类 号：V271.4[航空宇航科学与技术—飞行器设计]