Robust Networks for Initial Alignment of Strap-down Inertial Navigation Systems

摘要

This paper presents a method for the initial alignment of strap-down inertial navigation system on stationary base. In the coarse alignment, Magnetism compass and accelerometer are used to get the initial pose angles. In the precise alignment, radial basis function network (RBF network), in which radial function chosen by averaging method and weights adjusted by an improved Hi robust filter, is applied to correct the pose angles to ensure the robustness and real-time computing. Simulation results show that comparing with ordinary Hx filter learning algorithm and extended Kalman filter, this method is efficient and rapid for initial alignment