Scholarlink

<p>The tightly coupled integration (TCI) of GNSS Precise Point Positioning (PPP) and Micro-Electro-Mechanical Systems (MEMS) Inertial Navigation System (INS) has emerged as a promising approach for urban navigation, yet its performance degrades significantly by multipath interference common in complex urban environments. This study proposes a novel framework to enhance PPP/MEMS-INS integration through an adaptive code pseudorange classification and smoothing algorithm: (1) adaptive code pseudorange classification, where code measurements within a sliding window are statistically characterized and are used to classify subsequent data into normal/abnormal categories with adaptive weighting; (2) LSTM (Long Short-Term Memory)-based predictive modeling, which leverages initial data to train a LSTM network and smooths the abnormal code measurements of each satellite in challenging environments. Experimental results show that the proposed method reduces RMS errors in urban canyons by up to 48% and 42% in the horizontal and up directions respectively compared to traditional PPP/INS TCI.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Improved tightly-coupled PPP/MEMS-INS integration with adaptive code pseudorange weighting and LSTM-based smoothing for urban canyon navigation

  • 期刊论文

  • 发表日期：

    2025年11月3日

查看全文
  • Shoujian Zhang,
  • Dingxi Wang,
  • Lei Zhao,
  • Xingxing Li,
  • Haomeng Cui

摘要

The tightly coupled integration (TCI) of GNSS Precise Point Positioning (PPP) and Micro-Electro-Mechanical Systems (MEMS) Inertial Navigation System (INS) has emerged as a promising approach for urban navigation, yet its performance degrades significantly by multipath interference common in complex urban environments. This study proposes a novel framework to enhance PPP/MEMS-INS integration through an adaptive code pseudorange classification and smoothing algorithm: (1) adaptive code pseudorange classification, where code measurements within a sliding window are statistically characterized and are used to classify subsequent data into normal/abnormal categories with adaptive weighting; (2) LSTM (Long Short-Term Memory)-based predictive modeling, which leverages initial data to train a LSTM network and smooths the abnormal code measurements of each satellite in challenging environments. Experimental results show that the proposed method reduces RMS errors in urban canyons by up to 48% and 42% in the horizontal and up directions respectively compared to traditional PPP/INS TCI.