一种提高捷联惯性/里程计组合导航系统定位精度的后处理方法

测绘用捷联惯性/里程计组合导航系统多采用离线后处理技术.此类系统利用里程计位移微分获得的速度作为观测量,采用速度匹配,并通过待测路径中预置的Mark点校正航位推算的位置信息.本文建立了基于速度匹配的16维Kalman滤波模型,对全程采样数据进行正反向导航和滤波处理,以估计惯性器件和里程计的误差;在补偿相关误差后,再次进行正向导航和滤波解算,以获得更精确的姿态矩阵.随后,根据相邻两个Mark点之间的姿态信息进行航位推算,并更新里程计刻度系数、里程计与惯导间的姿态误差矩阵;若此Mark点间的位置误差过大,则重新进行航位推算以减小位置误差.结果表明,与传统正向滤波相比,采用该方法后系统的最大位置误差由1.72m降低到0.08m,定位精度提高了95％以上.     

快速定位定向系统设计及车载实验研究

定位定向系统是能为载体提供精确地理位置坐标、指北方向和姿态角的导航系统,通常用于舰船、飞机、车辆等功能平台,为平台上的设备提供准确的位置和姿态参考信息.本文针对车载平台机动性高的特点,设计能够实现运动中对准的快速定位定向系统,开展捷联惯导数字递推算法、航位推算、多源信息组合导航、动基座对准算法、系统免标定、误差补偿等算法和技术研究.最后,开展静态对准、静态导航和动态车载实验研究.实验结果表明,动态对准时间小于5min,对准姿态精度小于1mil,方位保持精度小于1mil/2h,横滚角、俯仰角保持精度小于0.5mil/2h,里程计/惯导组合水平定位精度小于0.15％D,卫星/惯导组合水平定位精度小于10m.     

智能航迹推算系统的研究

提出了一种新模糊滤波法用于补偿磁罗盘测量值中的低频干扰。在此基础上 ,进一步提出了一种新的滤波方法 :模糊 -补偿滤波法 ,以消除磁罗盘测量值中各种干扰的影响。仿真表明 :在较长的时间内 ,采用模糊 -补偿滤波法的航迹推算系统的航向估值误差和位置误差均较小。     

DR/GPS组合导航系统滤波技术研究

陆用航位推算系统（DR）的精度主要受到里程系数和航向误差的制约,其误差模型本质具有非线性,因此采用非线性滤波算法能显著提高里程系数和航向误差的估计精度。本文将粒子滤波应用到航位推算（DR）/GPS组合导航系统数据融合过程中,对航位推算 (DR) 与GPS组合导航系统中的里程系数误差和航向误差进行辨识估计,并对里程系数和航向进行修正。粒子滤波存在的主要问题是粒子的退化现象严重,本文将量子粒子群优化 (PSO) 算法与粒子滤波相结合,提出了量子PSO粒子滤波算法,该算法采用量子位对粒子进行编码,引入量子旋转门与变异操作保持了粒子集的多样性,通过量子PSO搜索寻优重新分配粒子,使粒子集有效地逼近真实的后验概率分布,从而有效地减轻了退化现象,提高了粒子滤波的精度。跑车实验结果表明,该算法有效地抑制了DR导航系统误差的增长,提高了组合导航系统的定位精度。     

小型无人机水平导航控制研究

在考虑小型无人机特点和多种导航方法的基础上,研究了利用ＧＰＳ／航路推算组合导航方法针对无人机进行水平导航控制的方案,并针对一种小型无尾无人机进行了导航控制规律设计,仿真结果表明,这种组合导航方法即简单又经济,精度高、实时性强,可靠性好,可以构成具有较强实用性的自主式导航系统。     

腰绑式行人导航系统基于缓存区的启发式航向补偿算法研究

腰绑式行人导航系统通过航位推算方法实现行人定位导航，对于实行灾难救援和单兵作战等任务十分便利。行人行进步长和航向的准确估计是实现航位推算的前提。根据腰绑式惯性传感器信号的特征，采用基于三轴合成加速度的峰值检测进行单步划分；利用步频和加速度方差信息构建线性步长模型。针对人员执行任务过程中由于运动方向不确定性导致传统启发式航向漂移消除算法过度修正甚至失效的问题，提出了一种基于缓存区的启发式航向补偿算法。根据运动方式选定N个复步作为一个缓存区，计算缓存区内相邻复步航向差值的方差，依据方差动态调整航向补偿算法的修正强度，避免过度修正。利用所提算法进行了矩形实验和操场实验，终点定位误差小于1%，表明了所提方法的有效性。     

陆用定位定向系统自主导航技术

由于车载自主导航精度受作战半径影响,长航时使用需要一定的保障条件,难于实现无依托发射的问题,提出了一种定向精度不受导航时间影响、定位精度不受作战半径影响的自主导航方案。通过分析惯性旋转调制导航、惯性/里程计组合原理及误差特性,采用罗经效应原理实现了高精度长航时自主定向;通过旋转调制导航抵消惯性器件误差的影响,利用航位推算隔离载体加速度和速度对罗经效应的影响,使航向误差完全可观,提升实时估计与修正精度。在此基础上引入了地图匹配技术进行自主定位,解决陆用定位精度与行驶里程相关的问题。仿真和试验结果表明,该技术采用地图信息辅助定位定向系统进行自主导航,在较低保障要求下,能够解决定向、定位误差积累问题,具有较强的理论意义和工程实用价值。     

基于PDR和磁场匹配的智能手机室内定位

目前，基于磁场的室内定位方法存在指纹采集无法快速建库的问题，导致匹配特征较少、无法快速进行匹配操作和航向角估计精度低。基于磁场梯度的快速近似，使用高斯牛顿迭代方法结合通过PDR测量的轨迹进行磁场轮廓匹配定位，提高了单点磁场指纹的可分辨性。基于步态模型的更新用作测量信息以改善C-INS的导航性能，提出了一种基于捷联惯性导航系统的行人航位推算算法。基于MFS参考位置，使用扩展Kalman滤波器控制PDR惯性导航的位置漂移误差，进一步提高了轨迹轮廓的精度。测试结果表明，该算法可以获得更好的位置估计和航向估计结果。     

Navigation and landing [A century of powered flight 1903-2003]

Navigation and landing for the first half century of aviation fundamentally relied on radio-based electronic aids. The military began serious investigation of radio navigation during World War I. Up until WW II map reading, dead reckoning, and various means of radio direction finding were the primary methods of determining aircraft position. Since WWII a number of other navigation techniques, such as Doppler radar, radio inertial, pure inertial, aided inertial, and today's integrated GPS inertial, have been developed. Most of the inertial systems evolved from the military ICBM and subsequent space programs.     

Road profile recognition for autonomous car navigation and Navstar GPS support

We discuss autonomous car navigation based on updating dead reckoning (DR) by road profile recognition (RPR). The navigation system requires sensors to detect changes in altitude and driving direction which are installed in modern cars for different purposes (e.g. ABS sensors). The layout of the navigation system is discussed and simulations are carried out over driving distances of approximately 150 km on the basis of realistic road data and ordinary sensor accuracies. Positioning errors of lower than 10 m (standard deviation) are observed. To achieve this accuracy the synchronization error between measured and mapped data must be continually estimated. The introduced navigation method is ideal to complete present commercial car navigation systems using Navstar GPS.     

基于量测预处理技术的GPS/DR组合算法研究

基于GPS在市区不能实现连续准确定位以及车辆航位推算系统(DR)的自主性,建立了GPS/DR组合的系统以及量测方程。以此为基础,根据GPS以及DR的采样频率的不同,引入量测预处理技术并用于GPS/DR的组合算法中,进而建立了采用量测预处理技术的卡尔曼滤波方程,最后通过实际的动态试验,论证了上述算法,并得出相应的结论。     

北斗／罗兰-C组合定位技术

“北斗一号”是我国的卫星导航系统,其有源定位模式使其很难得到广泛应用。为此,提出了一种北斗／罗兰-C组合定位方案以实现无源定位。该方案利用“北斗一号”两颗卫星和地面罗兰-C台站获得的距离差信息实现双曲线相交定位。仿真结果表明,该方案的精度可以满足一般定位要求,具有较强的实用性。     

Analysis of GPS and Loran-C performance for land vehicle navigationin the Canadian Rockies

Loran-C and GPS were assessed for vehicular navigation along selected roads of British Columbia during the winter of 1991. The general topography of this mountainous area is described, together with the specific topographic features and tree coverage characteristics of the 2000 km of roads tested on the mainland and on Vancouver Island. The configuration and characteristics of the Loran C Canadian West Coast chain along the roads used are described. The portable vehicle-mounted system used to collect and analyze the Loran-C and GPS signals along road profiles is described. The performance of Loran-C is analyzed in terms of signal to noise ratios (SNR), field strength, and time-difference distortions, as measured by differential GPS. These distortions, which can reach several hundred meters over distances of less than 20 km, are analyzed in terms of topographic features. The possibility of using these time-independent distortions to calibrate Loran-C for use along the above roads is discussed. Masking of GPS signals due to topographic features and tree coverage along the roads is analyzed. A comparative analysis of both Loran-C and GPS is presented in terms of signal availability and accuracy     

减小双轴旋转惯导中陀螺仪随机漂移影响的方法研究

针对双轴旋转惯导惯性器件的随机误差无法在导航过程中自动进行补偿的问题,提出优化的两位置重调(TPR)的方法来补偿系统的随机误差导致的方位和位置误差,以提高双轴旋转惯导的长航时精度.与传统两位置重调(CPR)方法相比,使用优化的误差传递方程的两位置重调的方法,可以在少于6h条件下估计出系统的方位误差,从而使得系统的位置精度和方位精度都得以极大的提高.根据惯性器件的随机误差导致的方位误差的特性,建立了TPR的误差模型.通过仿真,证明了该方法的有效性.     

Automobile Navigation: Where is it Going?

Automobile navigation systems based on dead reckoning, map matching, satellite positioning and other navigation technologies are under active development. ``Map intelligent' systems achieve high relative accuracy by matching dead-reckoned paths with road geometry encoded in a digital map data base that may also serve other functions such as vehicle routing and geocoding. Satellite-based navigation systems achieve high absolute accuracy but require dead reckoning augmentation because of signal aberrations in the automotive environment. Future systems will probably include multiple navigation technologies. Issues influencing the design of future systems include safety concerns regarding the driver interface, and the future availability of comprehensive map data bases, real-time traffic data, and mobile data communication links necessary for on-board generation of optimum routes.     

基于对偶四元数的INS/GPS组合导航算法研究

在基于对偶四元数的捷联惯导解算方法的基础上,推导了以惯性系作为导航系的惯导误差方程,在此基础上设计了卡尔曼滤波组合导航算法。通过激光惯导跑车采集数据,进行了仿真分析,试验结果表明,该组合导航算法能有效的消除惯导累积的速度误差和位置误差,相比于目前广泛应用的INS/GPS组合导航算法,本文描述了INS/GPS组合导航的另一种实现方式,获得了相当的精度,具有一定的工程应用价值。     

传统组合导航中的实用Kalman滤波技术评述

在随机线性系统建模准确的情况下,Kalman滤波是线性最小方差无偏估计。针对传统惯导/卫导组合导航的实际应用,难以精确建模,给出了常用的建模方法、状态量选取原则、离散化方法及滤波快速计算方法。讨论了平方根滤波、自适应滤波、联邦滤波和非线性滤波等技术的适用场合,并给出了使用建议。针对前人研究可观测度中未考虑随机系统噪声的缺陷,提出了更加合理的以初始状态均方误差阵为参考的可观测度定义和分析方法。提出了均方误差阵边界限制方法,可有效抑制滤波器的过度收敛和滤波发散。该讨论可为工程技术人员提供一些有实用价值的参考。     

The Effect of Aircraft Lateral Dynamics on Aircraft Positioning Accuracy When Using Loran-C

The dynamics of an aircraft following a fixed course line using Loran-C for position fixing are shown to interact with Loran-C receiver dynamics to result in cross-track aircraft positioning errors that are smaller than cross-track Loran receiver errors. In a particular case considered, this error reduction is on the order of 50 percent.     

全源定位与导航的统一理论框架构建

针对多导航传感器的信息融合复杂性的问题,提出了一种基于全源定位与导航的信息融合统一理论框架。该框架将目前导航领域的主要定位导航技术分成了与时间无关的定位导航技术方法以及与时间相关的定位导航技术。首先,主要从数学的角度将各种导航技术的测量模型抽象为统一的表达形式,提出了位置、姿态、速度函数的概念,讨论了各种测量方法的测量函数及其应用;其次,分析了航位、航姿、航速的推算方法;最后根据测量模型和运动模型构建了信息融合的基本方程,并讨论了基于贝叶斯估计的导航参数一般性估计方法。     

Quality control in integrated navigation systems

Real-time estimation of parameter sin dynamic systems, which becomes increasingly important in the field of high-precision navigation, requires real-time testing of the models underlying the navigation system. A real-time recursive testing procedure than can be used in conjunction with the Kalman filter algorithm is presented, along with diagnostic tools for inferring the detectability of particular model errors. The testing procedure consists of detection, identification, and adaptation. It can accommodate model errors in the measurement model and dynamic model of the integrated navigation system and is optimal in the uniformly-most-powerful-invariant sense