摘要
针对多传感器非线性系统中量测数据不能同时刻到达融合中心的情况,研究在无迹Kalman滤波(Unscented Kalman filtering,UKF)算法的基础上,按照量测数据到达的先后顺序进行序贯融合,提出了序贯无迹Kalman滤波(Sequential unscented Kalman filter,SUKF)算法。各个子系统通过无迹Kalman滤波器得到局部滤波估计,按照局部滤波结果到达融合中心的顺序,分别利用序贯协方差交叉融合(Sequential covariance intersection,SCI)算法和序贯逆协方差交叉融合(Sequential inverse covariance intersection,SICI)算法对各个子系统局部估计进行融合,避免了互协方差的计算,降低了计算负担。通过与集中式观测融合(Centralized observation fusion,CF)算法以及局部滤波器的精度对比分析,给出了3种融合算法的估计精度分析。用目标跟踪非线性系统仿真算例,验证了所提出的序贯滤波算法的有效性。
In view of the situation that the observation data in the multi-sensor nonlinear system cannot reach the fusion center at the same time,based on the unscented Kalman filtering(UKF)algorithm,the observation data are sequentially fused according to the order of data arrival to obtain a sequential unscented Kalman filter(SUKF)algorithm.The UKF is used to obtain the local filter estimates of each subsystem.Follow the order in which the local filtering results arrive at the fusion center,the sequential covariance intersection(SCI)algorithm and the sequential inverse covariance intersection(SICI)algorithm are used to fuse the local estimates of each subsystem,avoiding the calculation of cross-covariance and reducing the computation burden.By comparing the estimation accuracy with the centralized observation fusion algorithm and the local filter,the precision analysis of the three fusion algorithms is given.Simulation experiments on the target tracking nonlinear system verify the effectiveness of the algorithm.
作者
张雪楠
郑佰富
张润恒
刘志伟
高媛
ZHANG Xuenan;ZHENG Baifu;ZHANG Runheng;LIU Zhiwei;GAO Yuan(School of Electronic Engineering,Heilongjiang University,Harbin 150080,China)
出处
《黑龙江大学工程学报(中英俄文)》
2025年第1期37-45,共9页
Journal of Engineering of Heilongjiang University
基金
国家自然科学基金项目(61503125)
黑龙江省自然科学基金项目(Qc2013c62)
黑龙江省属本科高校2023年度“优秀青年教师基础研究支持计划”(YQJH2023139)
关键词
无迹Kalman滤波
序贯融合
序贯滤波
协方差交叉
逆协方差交叉
unscented Kalman filter
sequential fusion
sequential filter
covariance intersection
inverse covariance intersection
