摘要
针对目前低成本终端硬件质量较差,使得采集的数据受观测噪声、多路径效应、粗差等影响,导致质量较差且不稳定,利用常规算法难以实现高精度定位目的的问题,提出一种符合低成本终端特性的四分位法抗差自适应卡尔曼滤波方法:改进抗差自适应滤波模型,无须对新息向量进行标准化,通过采用四分位法动态确定阈值的方法建立抗差模型并求取方差膨胀因子;然后根据抗差权函数调整观测值的权重,剔除粗差并改善其定位精度.实验结果表明,小米 8终端在静态实时伪距差分(RTD)模式下东(E)、北(N)、天(U)方向精度分别提升 46.2%、51.9%和 50.2%;在静态实时动态差分(RTK)模式下 E、N、U方向精度分别提升 38.5%、33.3%和 57.5%;在行人动态 RTK模式下 E、N、U方向精度分别提升 56.6%、45.9%、44.9%;T5 终端在车载动态RTK模式下 E、N、U方向精度在 1 m以内,定位精度提升明显.
Abstract
Aiming at the problems that currently,the hardware quality of low-cost terminals is poor,which makes the collected data affected by observation noise,multi-path effect,gross error,etc.,leading to poor quality and lack of stability,and it is difficult to achieve the purpose of high-precision positioning by conventional algorithms,the paper proposed a quartile adaptive Kalman filtering method that meets the characteristics of a low-cost terminal:the adaptive filtering model was refined,without standardizing the new information vector,the model of resistance to error was established by using the quartile method to dynamically determinate the threshold,and the variance expansion factor was calculated;then,the weights of observations were adjusted according to the resistance weight function,the coarse error was eliminated,and the positioning accuracy was improved.Experimental results showed that in static real time differential(RTD)mode,the accuracy of Mi 8 terminal in E,N and U direction would be improved by 46.2%,51.9%and 50.2%,respectively;in the static real time kinematic(RTK)mode,the accuracy of E,N and U direction would be improved by 38.5%,33.3%and 57.5%,respectively;in the pedestrian dynamic RTK mode,the accuracy of E,N and U direction would be improved by 56.6%,45.9%and 44.9%,respectively;in the on-board dynamic RTK mode,the accuracy of the T5 terminal in the E,N and U directions would be less than 1 m,with significantly improved positioning accuracy.
NETL
NSTL
万方数据