马尔可夫矩阵修正IMM跟踪算法  

Research on adaptive Markov matrix IMM tracking algorithm

作　　者：封普文[1] 黄长强[1] 曹林平[1] 雍肖驹[1] 

机构地区：[1]空军工程大学航空航天工程学院,陕西西安710038

出　　处：《系统工程与电子技术》2013年第11期2269-2274,共6页Systems Engineering and Electronics

基　　金：航空科学基金(20105196016)资助课题

摘　　要：传统交互多模型(interacting multiple model,IMM)滤波算法中,马尔可夫概率转移矩阵参数固定,切换过程模型概率滞后。基于后验信息修正,扩展了一种在线更新马尔可夫概率转移矩阵的自适应跟踪算法,新算法克服了原算法只能交互2个模型的局限性。在计算过程中,依据不匹配模型误差压缩率的更新信息,在线调整先验马尔可夫概率转移矩阵,模型转换过程中更多地利用匹配模型的信息,而减小不匹配模型信息的影响,使收敛速度得到了提高。最后通过多模交互3个当前统计模型(current statistical model,CSM)验证了所提算法的有效性。The Markov parameter of the conventional IMM is a constant and it may be imposed to slow down the model switching. An adaptive Markov parameter IMM algorithm is developed based on modifying pos terior information. The new algorithm gets rid of the limit of the original algorithm just only using for two mod- els. The Markov transition probabilities can be modified during the process of online calculating by magnifying the matching model information and omitting the information of non-matching models simultaneously at the switching time. The algorithm accelerates system convergence speed. Finally, the validity of the new algorithm is proved through interacting three current statistical models.

关 键 词：交互式多模型 马尔可夫矩阵 后验信息 目标跟踪 “当前”统计模型 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]