Investigation of Real-Time Satellite-to-Satellite Tracking Using Extended Kalman Filters.

摘要

The feasibility of adapting an extended Kalman filter to the real-time satellite-to-satellite tracking problem is investigated. A filter using linear propagation and nonlinear measurements is formulated. This filter accepts tracker measurements to provide the state estimates (position, velocity, and acceleration) of the tracked satellite in the local inertial reference frame. Six different filters are programmed in American National Standards Institute (ANSI) FORTRAN. These filters employ various covariance update techniques and methods of processing the measurement vector. The programs are compared for computer program storage requirements, program execution speeds, and computational precision. The computer program storage requirements are determined by utilizing the FORTRAN IV-PLUS compiler on a PDP-11/45 minicomputer. The program execution speeds are found by timing on the PDP-11/45 the filter's reaction to a typical renedvous between two space vehicles.