Error Diagnosis in Space Navigation Integration Using Wavelet Multi-Resolution Analysis with General Regression Neural Network

Abstract

In this work, recent navigation systems rely on Kalman filtering for the fusion of data obtained from Global Positioning System (GPS) and the inertial navigation system (INS). The navigation combination offers consistent solutions of navigation by avoiding the generation of position errors with the consideration of time in case of INS. In current scenario, Kalman filtering INS/GPS method of integration contain some limitations related to stochastic error models of inertial sensors and the immunity to noise. The aim of this paper is to propose a system integration technique for the fusion of data from INS and GPS with the architecture of alternative general regression neural network (GRNN). The wavelet multi-resolution analysis (WMRA) is processed for comparing the position outputs of GPS and INS at various levels of resolution. The GR-ANN module is trained for predicting the position error of INS and offers vehicle positioning with increased accuracy.