New techniques for initial alignment of strapdown inertial navigation system

Some new techniques for initial alignment of strapdown inertial navigation system are proposed in this paper. A new solution for the precise azimuth alignment is given in detail. A new prefilter, which consists of an IIR filter and a Kalman filter using hidden Markov model, is designed to attenuate the influence of sensor noise and outer disturbance. Navigation algorithm in alignment is modified to feedback continuously for the closed-loop system. It is shown that the initial estimated variance setting of azimuth angle error can influence the speed of initial alignment significantly. At the beginning of alignment, Kalman filter must make a very conservative guess at the initial value of azimuth angle error to get a high convergent speed of the azimuth angle. It is pointed out that the low signal to noise ratio makes the ordinary setting of the estimated azimuth variance slow down the convergent speed of the azimuth angle. Also is shown that the large azimuth angle error problem can be solved well by our solution. The feasibility of these new techniques is verified by simulation and experiment.