Distributed model predictive control for low-thrust satellite
formation flying

Moustroufis, Dimitrios; Μουστρούφης, Δημήτριος

dc.contributor.author	Moustroufis, Dimitrios	en
dc.contributor.author	Μουστρούφης, Δημήτριος	el
dc.date.accessioned	2025-01-22T09:03:16Z
dc.date.available	2025-01-22T09:03:16Z
dc.identifier.uri	https://dspace.lib.ntua.gr/xmlui/handle/123456789/60904
dc.identifier.uri	http://dx.doi.org/10.26240/heal.ntua.28600
dc.rights	Αναφορά Δημιουργού - Παρόμοια Διανομή 3.0 Ελλάδα	*
dc.rights.uri	http://creativecommons.org/licenses/by-sa/3.0/gr/	*
dc.subject	Autonomous	en
dc.subject	Αυτόνομος	el
dc.subject	Gnc	en
dc.subject	Satellite formation	en
dc.subject	Distributed	en
dc.subject	Model predictive control	en
dc.subject	Κατανεμημένος	el
dc.subject	Προβλεπτικός έλεγχος	el
dc.subject	Σχηματισμός	el
dc.subject	Καθοδήγηση	el
dc.title	Distributed model predictive control for low-thrust satellite formation flying	en
dc.title	Κατανεμημένος προβλεπτικός έλεγχος για πτήσεις δορυφορικών σχηματισμών χαμηλής ώσης	el
heal.type	bachelorThesis
heal.classification	Spacecraft Guidance	en
heal.classification	Navigation and Control	en
heal.language	el
heal.language	en
heal.access	free
heal.recordProvider	ntua	el
heal.publicationDate	2024-07
heal.abstract	Satellite Formation Flying (SFF) is a new technology, which can offer unprecedented scientific capabilities, increased robustness, and reduced mission cost when compared to the use of traditional, monolithic spacecraft. The concurrent advancements in low-thrust propulsion methods (i.e. electric thrusters) enable SFF missions to be implemented with small platforms (such as nanosatellites) and for longer mission durations, but also enables new capabilities in orbital maneuver control accuracy and optimization. This work focuses on designing a Model Predictive Control (MPC) algorithm to serve the guidance and control functionalities for small formations of low-thrust capable satellites, while ensuring collision avoidance during reconfiguration maneuvers. The controller is implemented in a distributed fashion to reduce computational cost and avoid the single failure point present in a centralized architecture. At the outset of this work, a discussion on absolute and relative astrodynamics is carried out. The major state representations are introduced and models for the dominant sources of orbit perturbations in Low Earth Orbit (LEO) are provided. Then, the Relative Orbital Elements (ROE) state representation is introduced to model the relative motion between 2 spacecraft in proximity. Exploiting ROE, the differential equations describing the dynamics of the relative motion are presented in the form of a linear, time-varying system. Subsequently, the optimal control landscape is reviewed, and direct optimal control methods are presented in detail since a direct collocation method is used to formulate the optimal control problem of the MPC algorithm. Regarding MPC, it is situated among the other optimal control methods and its core aspects are discussed. Furthermore, an introduction to Distributed MPC (DMPC) is made, explaining its difference from Centralized and Decentralized formulations. The DMPC scheme used in this thesis assumes uncooperative agents, relying on predicted information from previous times to perform their update step simultaneously. Additionally, a dead reckoning approach is introduced which enables the formation members to perform their update step amid communication loss. With respect to the optimal control problem, the objective function and constraints are formulated in convex form, which enables a fast solution through effective solvers. In particular, the Ipopt solver is utilized, which exploits the constraints’ sparse structure to expedite the solution, with solution time measurement on a frequency-limited CPU proving amenability to on-board implementation. Focusing on the optimal control problem, the objective function, the constraints and the gradients of the constraints are derived. Moreover, the problem of recursive feasibility loss which arises from the distributed implementation is presented and methods are devised to mitigate it, namely: collision avoidance prioritization, a terminal relative velocity constraint, and an uncertainty tube around the predicted trajectories. The developed controller is then tested in various scenarios to verify performance on fuel expenditure and collision avoidance. To this end, fuel consumption on a benchmark problem is compared with other MPCs and a theoretically optimal impulsive control solution from the literature. Regarding collision avoidance, a 2-satellite position swap in an Along Track Formation (ATF) is carried out under various collision avoidance strategies. The best- performing strategy is then tuned and tested on a 2-satellite position swap in an ATF with a simultaneous relative orbit change of a third satellite. Next, the problem is expanded to a reconfiguration of a 4-deputy ATF to a Cart Wheel Formation (CWF). Finally, Monte-Carlo simulations are performed to test effectiveness of the uncertainty tube and the dead reckoning approach amid communication loss and absolute state determination uncertainty.	en
heal.advisorName	Παπαδόπουλος, Ευάγγελος	el
heal.committeeMemberName	Παπαδόπουλος, Ευάγγελος	el
heal.committeeMemberName	Πουλακάκης, Ιωάννης	el
heal.committeeMemberName	Αντωνιάδης, Ιωάννης	el
heal.academicPublisher	Εθνικό Μετσόβιο Πολυτεχνείο. Σχολή Μηχανολόγων Μηχανικών. Τομέας Μηχανολογικών Κατασκευών και Αυτομάτου Ελέγχου	el
heal.academicPublisherID	ntua
heal.numberOfPages	92 σ.	el
heal.fullTextAvailability	false