测绘学报 ›› 2021, Vol. 50 ›› Issue (4): 475-486.doi: 10.11947/j.AGCS.2021.20200258

• 大地测量学与导航 • 上一篇    下一篇

手机GNSS数据质量提取的变分模态分解法

朋子涵, 高成发, 刘永胜, 张瑞成, 尚睿   

  1. 东南大学交通学院, 江苏 南京 211189
  • 收稿日期:2020-06-24 修回日期:2021-02-05 发布日期:2021-04-28
  • 通讯作者: 高成发 E-mail:gaochfa@163.com
  • 作者简介:朋子涵(1996—),男,博士生,研究方向为手机GNSS精密定位。E-mail:pzhseu@163.com
  • 基金资助:
    国家自然科学基金(41974030);智能手机室内外一体化融合定位算法技术支撑服务(8521008666)

Variational mode decomposition method for estimation of GNSS data quality from a smartphone

PENG Zihan, GAO Chengfa, LIU Yongsheng, ZHANG Ruicheng, SHANG Rui   

  1. School of Transportation, Southeast University, Nanjing 211189, China
  • Received:2020-06-24 Revised:2021-02-05 Published:2021-04-28
  • Supported by:
    The National Natural Science Foundation of China (No. 41974030);The Indoor and Outdoor Integrated Fusion Positioning Algorithm Technical Support Service for Smartphone (No. 8521008666)

摘要: 观测噪声提取是数据质量分析与随机模型构建的基础。当前手机GNSS观测噪声提取主要采用三阶差分法与历元间差分法,这些方法的提取结果都会受到卫星相关性和历元间相关性的影响。本文提出结合变分模态分解(variational mode decomposition,VMD)提取手机观测噪声的方法。模拟分解试验表明:VMD能够较好地对混合信号进行分解。提取小米8手机GNSS观测噪声并进行分析,结果表明:GPS、BDS和Galileo 3系统伪距观测噪声计算结果一致,GLONASS系统伪距噪声大约是其他系统的两倍,四系统的载波观测噪声相当,手机GNSS观测噪声与Android系统版本无关。相比于高度角随机模型,载噪比随机模型更适用于手机GNSS定位。利用数据质量提取结果拟合载噪比随机模型,并进行定位试验。定位结果表明:相对于高度角随机模型,采用载噪比随机模型后手机伪距单点定位效果能提升25%以上。手机PPP平面定位结果能收敛至0.6 m以内,高程定位精度收敛至1.2 m以内。

关键词: 手机GNSS, 变分模态分解, 质量分析, 随机模型

Abstract: At present, the third-derivative approach or the inter-epoch difference method is mainly used to calculate the GNSS observation noise of the smartphone, but the results of these approaches will be affected by satellite correlation and time correlation. Aimed to the GNSS data noise estimation of smartphones, we propose an approach of noise estimation method based onvariational mode decomposition (VMD). Firstly, it uses VMD to extract low-frequency components of double-difference observations, then inversely calculates the single-difference observations of each satellite, and finally calculates the observation noise of each satellite. Simulation decomposition experiments show that VMD can decompose mixed signals satisfactorily. The analysis results of the GNSS data quality from the Xiaomi 8 indicate that the pseudorange noise of GPS, BDS and Galileo are analogous. The pseudorange noise of GLONASS is about twice that of the other systems, and the phase noise of the four systems are coincident. Observation noise will not be influenced by the operating system. In addition, the carrier-to-noise ratio stochastic model is more suitable for GNSS positioning of smartphones than elevation stochastic model. Compared with the using of latter model, the single point positioning effect of smartphones can be improved by more than 25% after using the carrier-to-noise ratio model. The PPP results of the smartphone can converge to within 0.6 and 1.2 m in horizontal and vertical components, respectively.

Key words: smartphone GNSS, variational mode decomposition, quality analysis, stochastic model

中图分类号: