基于决策树和SVD的INS/GPS/UWB联邦卡尔曼滤波算法  

An INS/GPS/UWB Federal Kalman Filter Algorithm Based on Decision-making Tree and SVD

作　　者：时潮 孙飞 梅子杰 罗晨 关翔中 刘乔 SHI Chao;SUN Fei;MEI Zijie;LUO Chen;GUAN Xiangzhong;LIU Qiao(Shanghai Electro-Mechanical Engineering Institute,Shanghai 201109,China)

机构地区：[1]上海机电工程研究所,上海201109

出　　处：《无线电工程》2025年第3期566-571,共6页Radio Engineering

摘　　要：针对面向地面无人平台(Unmanned Ground Vehicle,UGV)的无人机(Unmanned Aerial Vehicle,UAV)动态跟踪和自主导航的实现问题,通过采用决策树算法建立信息源分类判据,提出基于奇异值分解(Singular Value Decomposition,SVD)的联邦卡尔曼滤波算法,构建了一种组合导航系统,利用全球定位系统(Global Position System,GPS)和超宽带(Ultra WideBand,UWB)技术实时提供UAV相对于无人车的位置信息,修正补偿惯性导航系统(Inertial Navigation System,INS)的误差,提高系统的可靠性和容错性。仿真实验表明,该联邦卡尔曼滤波导航系统能够满足UAV在GPS信号或UWB信号无效情况下定位导航的精度要求,可大大提高INS/GPS/UWB组合导航系统的定位精度及容错能力。To deal with the problem of dynamic tracking and autonomous navigation for Unmanned Aerial Vehicle(UAV)that track Unmanned Ground Vehicle(UGV),the classification criteria about information source is established by decision-making tree,and a federal Kalman filter integrated navigation system based on Singular Value Decomposition(SVD)is established.Global Position System(GPS)and Ultra Wide-Band(UWB)technology are used to provide in real time the information about the position of the UAV relative to the UGV and correct the errors of the Inertial Navigation System(INS).Simulation results show that the UAV can obtain reliable navigation information by this modified federal Kalman filter algorithm even when the UWB or GPS signal is continuously interrupted for a period and the INS/GPS/UWB navigation system greatly improves the positioning accuracy and fault tolerance of the system.

关 键 词：决策树 修正补偿惯性导航系统/全球定位系统/超宽带技术 联邦卡尔曼滤波 奇异值分解 

分 类 号：TN92[电子电信—通信与信息系统]