The Application of Recurrent Neural Networks in the GPS Vehicle Navigation Positioning Prediction

摘要

This paper applies recurrent neural networks in the GPS vehicle navigation positioning prediction. A technique of predicting the vehicle positioning information based on the recurrent neural networks with the GPS receiver losing the GPS positioning signals is presented in this paper. The dynamic backpropagation algorithm is used to train the diagonal recurrent neural networks in the pattern of tracking study to predict the future vehicle positions in a limited period. Numerical training and studying examples show that the diagonal recurrent neural networks can give relatively accurate and successive predictions about the future vehicle positions when the receiver loses the GPS signals in a limited period.