Strapdown inertial system alignment using statistical filters - A simplified formulation.

Abstract

Introduction I of Strapdown inertial systems requires determining the attitude (direction cosine) matrix which relates a vehicle-fixed set of axes to a reference or navigation coordinate frame. If the vehicle motion is measured in the reference frame by some auxiliary means such as another set of inertial sensors or doppler radar, transfer alignment can be accomplished by comparing similar quantities measured by the Strapdown system gyros and accelerometers. Several techniques to accomplish transfer alignment exist, including gyrocompassing and the use of Kalman-like (recursive) filters. Typically, alignment schemes employing the Kalman optimum technique involve many state variables to describe relative motion between the vehicle and reference axes. While such filters give generally better accuracy and faster alignment than gyrocompassing they impose a heavy computation burden on the alignment system. In particular, they require solution of a large number of simultaneous differential equations with coefficients dependent upon vehicle motion. The formulation of the Strapdown system transfer alignment problem developed in this paper requires only three state variables which do not vary with time. Consequently, a good deal of the computer burden imposed by other recursive filter implementations is eliminated.