RBFNN Aided Extended Kalman Filter for MEMS AHRS/GPS

Abstract

A Radial Basis Function Neural Network (RBFNN)-aided Extended Kalman Filter (EKF) is designed towards a low cost solid-state integrated navigation system. This system incorporates measurements from an attitude and heading reference system (AHRS) and a GPS, providing unaided, complete and accurate navigation information for land vehicles. To realize the EKF algorithm, the architectures of this AHRS/GPS and the description of Pseudo_range- Pseudo_range Rate -Heading measurements model are intensively illustrated. In sequence, the fundamentals of radial basis function (RBF) technique are discussed by the procedure of aiding mode and realization process. The simulation test shows when the carrier is in dynamic environment, the navigation parameters are relatively precise, even if the accuracy of the sensors is modest. This fusion filter approach, illustrated here proves to be a practical approach for navigation parameters estimation in real time.