Prediction of DGPS corrections with neural networks

Abstract

Applies neural network modelling to predicting DGPS (Differential Global Positioning System) corrections. The paper first briefly introduces GPS and DGPS navigation principles and aircraft navigation performance requirements. Following a discussion of the temporal characteristics of the DGPS corrections, a technique for predicting the DGPS corrections based on diagonal recurrent neural network (DRNN) modelling is presented. Numerical examples show that the prediction accuracy is better than 1 m for 10 s prediction and 1.3 m for 30 s prediction, respectively, which can maintain the aircraft navigation at the required accuracy for a period of 30 s.