The possibilities of using the Kalman filter for cleaning (filtering) signals from gy-roscopes and accelerometers of a strapdown inertial navigation system (SINS) with a rough initial alighnment on a fixed base are considered. For this, algorithms for averaging filters with forward and backward computations (forward and backward filters) are proposed. It is shown that with a relatively short recording of the signal (30 s), in comparison with the simple average of the signal, it is possible to

more »

... gnifi-cantly, practically to zero, reduce the influence of the measurement noise. This is achieved by successively applying forward and then backward filters. The results were obtained both in the modeling of filtration and in the field experiment of the initial alighnment.