Navigation observation target selection method based on Q learning

Abstract

The invention discloses a navigation observation target selection method based on Q learning. Q learning state spaces are set for different observation areas, a Q function is calculated based on measurement innovation of a navigation filter, and an observation area of a sensor is selected according to the Q function; further, for each alternative space target, a target evaluation function is established by using a statistical value of the measurement innovation, and a space target for navigation is selected in the observation area according to the target evaluation function; as the learning process proceeds, the sensor will adaptively select a target that contributes to improving the filtering estimation accuracy for observation. The main technical content of the patent can be used in an autonomous navigation system for measurement in a space target sight line direction . The method can achieve the optimal selection of an observation target under the condition that the prior information of a part of space targets is not accurate to guarantee the autonomous navigation precision of a spacecraft, so that the improvement of the capability of a navigation system for dealing with the uncertainty of a measurement model is improved, and the adaptability of the space system to the environment change is enhanced.

Description

technical field [0001] The invention belongs to the technical field of spacecraft autonomous navigation. Background technique [0002] In the spacecraft autonomous navigation system based on line-of-sight (LOS) direction measurement, the star camera installed on the spacecraft is used as a navigation sensor to obtain the unit line-of-sight vector information of known space targets, and then, combined with the orbital dynamics model, through The Extended Kalman Filter (EKF) processes the LOS direction observations to estimate the absolute position and velocity vector of the spacecraft. [0003] Navigation observation target selection is an important link in spacecraft autonomous navigation based on line-of-sight measurement. The star camera on the spacecraft obtains the line-of-sight observations of space objects in the inertial system through photographic observation of space objects and background stars. Limited by the field of view of the star camera (such as 1°×1°), onl...

AI Technical Summary

Problems solved by technology

[0006] The technical problem solved by the present invention is: to overcome the deficiencies of the prior art, provide a method for selecting navigation observation targets based on Q-learning, and solve the problem that the uncertainty of the measurement model caused by the inaccurate prior knowledge of space targets affects the performance of the navigation system , which can adaptively complete the optimal selection of observation targets and improve the performance of navigation filters

Images