摘要
针对传统的恒星日滤波方法难以抑制GNSS多星座组合观测的多路径误差问题,该文提出了一种用于削弱多路径误差的卷积神经网络和门控循环单元的融合模型(CNN-GRU),通过挖掘GNSS多路径误差的深层变化规律,对短基线中的多路径误差进行实时预测和改正。实验结果表明,该模型对于BDS和不同系统间组合的多路径误差可以起到实时削弱的作用,能够显著提高定位精度。此外,CNN-GRU模型的预测结果具有良好的鲁棒性,并且不受粗差的影响。
Aiming at the problem that the traditional stellar day filtering method is difficult to suppress the multipath error of GNSS multi-constellation combination observation,this paper proposes a fusion model of convolutional neural network(CNN)and gated recurrent unit(GRU)for weakening multipath error(CNN-GRU).By mining the deep variation law of GNSS multipath error,the multipath error in short baseline is predicted and corrected in real time.The experimental results show that the model can weaken the multipath error between BDS and different systems in real time,and can significantly improve the positioning accuracy.In addition,the prediction results of CNN-GRU model have good robustness and are not affected by gross errors.
作者
刘春阳
刘超
陶远
徐胜华
樊亚
LIU Chunyang;LIU Chao;TAO Yuan;XU Shenghua;FAN Ya(Anhui University of Science and Technology,Huainan,Anhui 232001,China;China University of Mining and Technology,Xuzhou,Jiangsu 221116,China;Chinese Academy of Surveying and Mapping,Beijing 100036,China;China Building Materials Industry Geologic Exploration Center Guizhou Groups,Guiyang 551400,China)
出处
《测绘科学》
CSCD
北大核心
2022年第11期40-47,72,共9页
Science of Surveying and Mapping
基金
安徽省自然科学基金面上项目(2108085MD130)
安徽省高校自然科学研究重点项目(KJ2020A0312)
作者简介
刘春阳(1991—),男,安徽六安人,讲师,博士,主要研究方向为时空数据分析、GNSS数据处理。E-mail:cyliu6666@163.com;通信作者:刘超,副教授,E-mail:chliu1@aust.edu.cn
