Fast initial alignment for inertial navigation system based on high order sliding mode observer and Kalman filter

Abstract

The inertial navigation system is a dead reckoning system, thus initial alignment for an inertial navigation system plays an important role in the accuracy of it. In this paper, a novel approach for initial alignment in an inertial navigation system with increased speed and accuracy is proposed. This method has two stages, which integrates the Kalman filter and a high order sliding mode observer. In the inertial navigation system, leveling misalignment angles reach the steady-state faster than the azimuth misalignment angle does, which means the azimuth alignment takes a considerable time for initial alignment. Therefore, in this paper at the first stage estimations of state variables of the system are obtained using the Kalman filter and whenever all variables (except azimuth alignment) reach steady-state, the second stage begins. In the second stage, the estimation which is obtained by the Kalman filter is used as the input to design an equivalent system with unknown inputs for inertial navigation system. A high-order sliding mode observer is then used to estimate the states of a system with an unknown input for estimating the azimuth alignment angle. This method not only increases the speed of estimation but also has comparable accuracy.