Digital Prefiltering of Correlated Measurement Noise Using Regression Lines

摘要

The fast initial fine alignment of an inertial platform with a Kalman filter accelerated by a digital prefilter is shown. The digital prefilter is used to estimate certain state variables by constructing linear mean square regression lines from each batch of measurements with colored noise. The developed prefilter is characterized by a brief computation time and small storage requirements, which makes a very high data rate possible. The choice of filter structure and parameters is discussed referring to simulation results. The filter algorithm is illustrated step by step, both on line and off line. It is compared to prefiltering by averaging and is shown to perform well.