基于多层感知器神经网络的单频观测伪距精度优化研究

打开文本图片集

中图分类号：TP273；TP242 文献标识码：A 文章编号：2096-4706（2025）17-0038-07

Abstract：Aimingat the problem that the positioningaccuracyof single-frequency receiveris limited dueto ionospheric delay，the Multi-Layer Perceptron （MLP）neural network isused to predicttheionospheric delay to improve the pseudo-range accuracyof single-frequency observation.Theobservation dataand broadcast ephemeris files ofIGS data centerof Wuhan Universityarecoected，andtheMPneuralnetworkmodelisonstructedafterpreprocessingThemodeltakes thepsedrange ofsingle-frequencyobservationanditsrelatedcharacteristicsasinput，andtheionosphericdelaycorrctedbytheobseation data ofthedual-frequencyreceiveras thepredictiontarget.Theexperimentalresultsshowthat he averagepredictionaccuracyof the model on the test set is 81.71% ，which is significantly beter than the traditional Klobuchar model （the average compensation accuracy is about 50%～60% ）

Keywords： single-frequency observed pseudo-range;Multi-Layer Perceptron neural network; ionospheric delay; Klobuchar model; accuracy optimization

0 引言

在GPS定位技术中，伪距观测精度对定位结果的准确度起着关键作用。（剩余11225字）