KALMAN FILTER INERTIAL NAVIGATION SYSTEM ERROR MODEL BASED ON FILTER STABILITY CONSIDERATIONS

摘要

Two forms of Kalman filter system error models are evaluated using filter stability considerations. These error models are based on one of two assumptions used to obtain reduced state models for wander azimuth navigation system mechanizations; either the navigation frame azimuth alignment, δθ_z, or the local azimuth tilt error with respect to the navigation frame, φ_z, is zero. The stability evaluation assumes that body referenced Doppler velocity and navigation frame GPS position measurements are used. Stability and monte carlo simulation results show that with the combination of Doppler and GPS, the latter assumption produces the more stable and accurate filter estimates.