模型参数自适应的低复杂度ATPM-VSIMM算法  

Low-complexity ATPM-VSIMM algorithm with adaptive model parameters

作　　者：曾浩[1] 母王强 蒋阳[1] 杨顺平 ZENG Hao;MU Wangqiang;JIANG Yang;YANG Shunping(School of Microelectronics and Communication Engineering,Chongqing University,Chongqing 400044,China;General Technology Department,Southwest Electronic Technology Research Institute,Chengdu 610036,China)

机构地区：[1]重庆大学微电子与通信工程学院,重庆400044 [2]西南电子技术研究所共性技术部,四川成都610036

出　　处：《通信学报》2023年第9期25-35,共11页Journal on Communications

基　　金：CETC联合基金资助项目(No.629010204)。

摘　　要：在机动目标跟踪中,针对交互式多模型算法使用固定模型集和固定转移概率矩阵导致跟踪精度下降的问题,提出模型参数自适应更新的低复杂度ATPM-VSIMM算法。所提算法根据系统新息变化情况来判断目标是否出现机动,从而调整模型集的状态噪声,实现模型集的自适应更新;然后,根据模型后验概率变化情况和模型间的相互切换关系,准确地计算出转移概率矩阵,从而提高系统运动模型和目标运动轨迹的匹配程度,保证跟踪系统具有滤波精度高和响应速度快的优点。从模型后验概率初值、转移概率矩阵初值和状态噪声三方面验证了所提算法的有效性。仿真结果表明,ATPM-VSIMM算法的空间位置跟踪精度比现有算法提高了8%左右。Aiming at the problem that for maneuvering target tracking,the accuracy of tracking degraded in interacting multiple model algorithms due to the fixed model sets and the fixed transition probability matrix,a low-complexity ATPM-VSIMM algorithm was proposed,which could update the model parameters adaptively.The maneuvering situation of the target was judged according to the innovation changes of the system,and the state noise of the model sets was adjusted to realize the adaptive update of the model sets.Then,the more accurate transition probability matrix was computed through the change of the model posterior probability and the inter-model switching relationship.Therefore,the matching degree between the system motion model and the target motion trajectory was improved.Finally,the high filtering accuracy and the fast response speed of the tracking system were guaranteed.The effectiveness of the proposed algorithm was verified through three aspects that are the initial value of the model posterior probability,the initial value of the transition probability matrix,and the state noise.Simulation results demonstrate that the filtering accuracy of the ATPM-VSIMM algorithm is improved about 8%compared with the existing algorithms.

关 键 词：机动目标跟踪 自适应状态噪声协方差矩阵 自适应转移概率矩阵 变结构交互式多模型 

分 类 号：TN953[电子电信—信号与信息处理]