GNSS Multipath Error Reduction based on Time Convolutional Network
Multipath error is the main error source of the short baseline of Global Navigation Satellite System(GNSS).Considering the difficulty to obtain the revisit period of the classical Sidereal Filte-ring(SF)satellite under the GNSS multi-system condition and the problem that the Multipath Hem-ispherical Map(MHM)requires a large amount of data support,this paper introduces the Time Convolutional Network(TCN)model to make full use of its deep mining capability of massive data to establish the multipath error model in GNSS coordinate sequences,and to perform multipath the real-time weakening of the errors.Through 15 days of continuous experimental data testing,the re-sults showed that the new method overcame the difficulty of applying the classical SF method to the weakening of multi-system GNSS multi-path errors,and required only one day of training samples to realize the real-time weakening of multi-path errors in subsequent days,and its weakening effect was significantly better than that of MHM model,showing good generalization ability.
Global Navigation Satellite Systemmultipath errorTime Convolutional NetworkSide-real FilteringMultipath Hemispherical Map
万方数据
维普
