INTRODUCTION TO DATA FUSION AND APPLICATIONS IN ASTRODYNAMICS

摘要

This paper reviews the precepts of multi-sensor data fusion and shows how they are applied to modern astrodynamics. Of necessity terrestrial sensor systems are sparse and diverse. Measurements include angles only optical (telescope) observations, state of motion measurements with radars, occasional capability to range on the communications links to the satellites, and (rarely) satellite position information from onboard Global Positioning System (GPS) transceivers. The hypotheses that are tested against observations are the equations of motion augmented with assumptions about the nonuniformity of the Earth's geopotential, the influence of atmospheric drag, the existence of light pressure, the influence of other heavenly bodies, and even the slow dynamics of the oceans and the Earth's mantle and core. If we model these phenomena well, we can estimate satellite orbits relatively accurately. More important, we can sustain accurate estimates with few observations. If we model these phenomena poorly, variances from measurements will be large, orbit uncertainty will be large, and many more observations might be necessary. Astrodynamics employs a rich spectrum of data fusion techniques. Above all we strive to maintain mathematical and physical rigor so that anomalies are traceable to physical or mathematical processes amenable to modification.