一种复杂环境下UWB测距误差预测方法  

A UWB ranging error prediction method in complex environments

作　　者：李圣令 刘克中[1,2,3] 陈聪 王一飞 王国宇 陈默子 郑凯 LI Shengling;LIU Kezhong;CHEN Cong;WANG Yifei;WANG Guoyu;CHEN Mozi;ZHENG Kai(School of Navigation,Wuhan University of Technology,Wuhan 430063,China;Hubei Key Laboratory of Inland Shipping Technology,Wuhan 430063,China;National Engineering Research Center for Water Transport Safety,Wuhan 430063,China;CSSC Cruise Technology Development Co.,Ltd.,Shanghai 200137,China)

机构地区：[1]武汉理工大学航运学院,武汉430063 [2]内河航运技术湖北省重点实验室,武汉430063 [3]国家水运安全工程技术研究中心,武汉430063 [4]中船邮轮科技发展有限公司,上海200137

出　　处：《导航定位学报》2024年第1期85-96,共12页Journal of Navigation and Positioning

基　　金：国家自然科学基金面上项目(51979216);湖北省自然科学基金创新群体项目(2021CFA001);湖北省自然科学基金青年项目(20221J0059)。

摘　　要：针对复杂室内环境中,多径干扰和视距遮挡影响室内定位精度的问题,提出一种复杂环境下UWB测距误差的预测方法:根据超宽带(UWB)室内定位方法抗干扰能力强、定位精度高的特点,从理论上分析影响超宽带信号测距精度的因素;然后通过模型量化分析每种影响因素对测距误差的影响程度,改进以往通过信道脉冲响应特征分析单一的视距遮挡参数来提高定位精度的补偿算法;最后从数据挖掘和机器学习的角度出发,将多径效应、视距遮挡、标签与基站的距离、标签移动速度、标签与基站天线俯仰角以及气象因素(温度、湿度、压强)等对测距精度造成影响的重要因素作为特征进行分类和归类,并使用梯度提升决策树模型对数据集进行训练和预测。实验结果表明,该模型可以根据当前状态下的组合特征值估计出测距误差值,将当前预测的误差值补偿到测量值上,可以有效提高超宽带室内定位的精度。Aiming at the problem that multipath interference and line of sight occlusion affect indoor positioning accuracy in complex indoor environments,the paper proposed an ultra-wide band(UWB)ranging error prediction method in complex environments:based on the advantages of strong anti-interference ability and high positioning accuracy of UWB indoor positioning,the factors that affect the ranging accuracy of UWB signals were theoretically analyzed;then the impact of each influencing factor on the ranging error was quantitatively analyzed through a model in order to improve the compensation algorithm for upgrading the positioning accuracy by analyzing a single line of sight occlusion parameter through channel pulse response characteristics in the past;finally,from the perspective of data mining and machine learning,the important factors including multipath effects,line of sight occlusion,distance between tags and base stations,tag movement speed,tag and base station antenna pitch angle,and meteorological factors(temperature,humidity,pressure)that affect ranging accuracy were classified and categorized as features,and a gradient boosting decision tree model was used to train and predict the dataset.Experimental result showed that the proposed model would estimate the ranging error value based on the combined feature values in the current state,and compensate the current predicted error value to the measured value,which could effectively improve the accuracy of UWB indoor positioning.

关 键 词：室内定位 超宽带(UWB)定位 测距误差 梯度提升决策树 机器学习 

分 类 号：P228.9[天文地球—大地测量学与测量工程]