Neuroadaptive observer design for spacecraft attitude control and formation attitude synchronization

Abstract

Spacecraft attitude-tracking control problems require uncertainties and disturbances- rejecting controllers to perform well in practical scenarios. The objective of this research is to design an adaptive controller that uses Modi ed State Observer (MSO) to improve upon the controller's performance for spacecraft attitude-tracking. The application scope of the MSO-based control strategy is then extended for attitude synchronization problems. In this thesis, a spacecraft is considered to be operating under in uence of unmod- eled system dynamics and external disturbances. The e ects of these factors are modeled as an unknown angular acceleration function, which is predicted by using an observer that depends on a neural network. The rst part of this research is focused on augmenting a classical inverse controller with an MSO observer (i.e. unknown function estimating ob- server). Chebyshev neural network is employed in designing the MSO such that the unknown function prediction performance is enhanced, which consequently enhances attitude-tracking performance. To validate the controller's enhanced performance, two numerical examples are presented to highlight the di erences in performance of the proposed versus existing con- trollers. In the second part of the research, the proposed attitude controller is augmented to solve attitude synchronization and stabilization of spacecraft in a leaderless formation- ying network. The communication time-delay, which is inherent and unpreventable, is considered in the course of designing attitude-synchronizing controller. The proposed methodology em- ploys a graph theory-based model for the communication network to determine a desired angular velocity pro le, which guarantees angular velocity stabilization and attitude consen- sus among spacecraft. The MSO-based inverse controller drives the angular velocity error and attitude error to zero such that the desired reference pro le is attained. Two numeri- cal examples are presented to validate that the controller is able to reject disturbances and uncertainties while achieving attitude synchronization.