Initial alignment of Inertial Navigation System based on a predictive iterated Kalman filter

Abstract

Inertial navigation systems (INSs) are widely used in practical applications. This paper considers the problem of initial alignment for an inertial navigation system. As the modeling error of an INS is difficult to be measured accurately, it will decrease the accuracy of the initial alignment. In this paper, a predictive iterated Kalman filter (PIKF) is proposed for a class of INSs with the modeling error and Gaussian noise. Firstly, the modeling error is estimated by a predictive filter. Then, in order to reduce the error between the one step prediction of Kalman filter and the true value, a new filtering method combining a predictive filter with an iterative Kalman filter is used to improve the alignment accuracy. Finally, simulations for the stationary alignment of an INS are given to show the efficiency of the proposed approach.