Future space missions increasingly demand efficient and reliable control strategies for executing proximity operations, while the widespread adoption of small satellites, such as CubeSats, introduces additional constraints on the formulation of control laws. This thesis addresses direction-constrained low-thrust optimal control problems motivated by an underactuated case, where thrust can only be applied along a single body-fixed axis. The study focuses on two representative proximity operations missions: close-range rendezvous and relative hovering. Efficient solution strategies are proposed to handle both energy-optimal and fuel-optimal formulations, beginning with analytical solutions for energy-optimal cases that serve as high-quality initial guesses for solving the more computationally demanding fuel-optimal problems. The work starts by formulating unconstrained low-thrust optimal control problems and solving them using both analytical techniques based on the state transition matrix and numerical continuation methods. A control-duration selection algorithm is developed to ensure fuel efficiency in low-thrust missions. Additionally, a long-term hovering strategy is proposed, maximizing hovering time while minimizing fuel usage by exploiting naturally periodic relative orbits and a fast target-point selection method. The research is then extended to direction-constrained scenarios where thrust is limited to predefined directions, for example, tangential, radial, or normal. Energy-optimal solutions in these configurations are derived analytically, while fuel-optimal solutions are obtained through a semi-analytical Non-Linear Programming (NLP) process, initialized by the energy-optimal result. In particular, the tangential case benefits from an asymptotic analytical propagation technique, enabling fast solution generation. These predefined direction solutions are further used to formulate a two-phase control strategy to achieve 6-degree-of-freedom orbital control. The thrust is constrained to a single predefined direction during each phase, one dedicated to in-plane motion control and the other to out-of-plane motion control. In addition, a user-defined intermediate coast phase is incorporated between the thrust phases, enabling reorientation of the thrust direction and accommodating supplementary mission constraints. The study is further expanded to address control problems involving constant thrust directions defined in both the local radial–transverse–normal frame and the Earth-centered inertial frame. A least-square method is employed to identify candidate thrust directions based on energy-optimal solutions, which subsequently serve as initial guesses for the NLP procedure to obtain fuel-optimal solutions. Several state-propagation schemes are developed and compared to assess their trade-offs in terms of computational cost and solution fidelity. An efficient overall strategy is proposed to determine the optimal constant control direction along with its corresponding optimal bang-off-bang thrust profile. The effectiveness and robustness of proposed algorithms are validated through comprehensive numerical simulations and Monte Carlo analyses. The results demonstrate that the developed methods provide accurate, computationally efficient, and easily implementable solutions for proximity operations involving direction-constrained low-thrust control. These features make the approaches particularly well-suited for onboard implementation in CubeSat missions and small spacecraft requiring autonomous guidance with limited actuation authority.
Le future missioni spaziali richiedono sempre più strategie di controllo efficienti e affidabili per l’esecuzione di operazioni di prossimità, mentre la diffusione dei piccoli satelliti, come i CubeSat, introduce ulteriori vincoli nella formulazione delle leggi di controllo. Questa tesi affronta problemi di controllo ottimo a bassa spinta con vincoli direzionali, motivati da un caso sottoattuato in cui la spinta può essere applicata soltanto lungo un singolo asse fissato al corpo. Lo studio si concentra su due missioni rappresentative di operazioni di prossimità: il rendezvous a corto raggio e lo hovering relativo. Vengono proposte strategie di soluzione efficienti in grado di gestire formulazioni sia a energia ottima sia a combustibile ottimo, a partire da soluzioni analitiche per i casi a energia ottima che fungono da stime iniziali di alta qualità per la risoluzione dei problemi a combustibile ottimo, più onerosi dal punto di vista computazionale. Il lavoro inizia formulando problemi di controllo ottimo a bassa spinta senza vincoli e risolvendoli mediante tecniche analitiche basate sulla matrice di transizione di stato e metodi di continuazione numerica. Viene sviluppato un algoritmo per la selezione della durata del controllo, allo scopo di garantire efficienza nel consumo di carburante nelle missioni a bassa spinta. Inoltre, viene proposta una strategia di hovering di lungo periodo, che massimizza il tempo di stazionamento riducendo al minimo il consumo di carburante, sfruttando orbite relative naturalmente periodiche e un rapido metodo di selezione del punto obiettivo. La ricerca viene poi estesa a scenari vincolati nella direzione della spinta, limitata a direzioni predefinite, per esempio tangenziale, radiale o normale. Le soluzioni a energia ottima in queste configurazioni sono ricavate analiticamente, mentre quelle a combustibile ottimo sono ottenute mediante un processo semi-analitico di Programmazione Non Lineare (NLP), inizializzato tramite le soluzioni a energia ottima. In particolare, il caso tangenziale beneficia di una tecnica asintotica di propagazione analitica, che consente una generazione rapida delle soluzioni. Le soluzioni con direzione predefinita vengono inoltre utilizzate per formulare una strategia di controllo a due fasi per ottenere il controllo orbitale a 6 gradi di libertà. La spinta è vincolata a una direzione predefinita durante ciascuna fase: una dedicata al controllo del moto complanare e l’altra al controllo del moto fuori piano. Viene anche introdotta una fase intermedia di coasting definita dall’utente, che permette la riorientazione della direzione di spinta e l’integrazione di vincoli aggiuntivi della missione. Lo studio viene ulteriormente ampliato per affrontare problemi di controllo con direzioni di spinta costanti, definite sia nel sistema locale radiale-trasversale-normale sia nel sistema inerziale geocentrico. Un metodo ai minimi quadrati è utilizzato per identificare direzioni di spinta candidate sulla base delle soluzioni a energia ottima, le quali fungono successivamente da stime iniziali per la procedura NLP per ottenere soluzioni a combustibile ottimo. Vengono sviluppati e confrontati diversi schemi di propagazione dello stato per valutarne i compromessi in termini di costo computazionale e accuratezza della soluzione. Si propone quindi una strategia globale efficiente per determinare la direzione di controllo costante ottimale insieme al corrispondente profilo di spinta bang-off-bang ottimale. L’efficacia e la robustezza degli algoritmi proposti sono validate tramite simulazioni numeriche approfondite e analisi Monte Carlo. I risultati dimostrano che i metodi sviluppati forniscono soluzioni accurate, computazionalmente efficienti e facilmente implementabili per operazioni di prossimità che implicano controllo a bassa spinta con vincoli direzionali. Queste caratteristiche rendono gli approcci particolarmente adatti all’implementazione a bordo di missioni CubeSat e piccoli veicoli spaziali che richiedono guida autonoma con capacità di attuazione limitate.
Direction-constrained low-thrust optimal control for spacecraft proximity operations
Zhao, Chuncheng
2025/2026
Abstract
Future space missions increasingly demand efficient and reliable control strategies for executing proximity operations, while the widespread adoption of small satellites, such as CubeSats, introduces additional constraints on the formulation of control laws. This thesis addresses direction-constrained low-thrust optimal control problems motivated by an underactuated case, where thrust can only be applied along a single body-fixed axis. The study focuses on two representative proximity operations missions: close-range rendezvous and relative hovering. Efficient solution strategies are proposed to handle both energy-optimal and fuel-optimal formulations, beginning with analytical solutions for energy-optimal cases that serve as high-quality initial guesses for solving the more computationally demanding fuel-optimal problems. The work starts by formulating unconstrained low-thrust optimal control problems and solving them using both analytical techniques based on the state transition matrix and numerical continuation methods. A control-duration selection algorithm is developed to ensure fuel efficiency in low-thrust missions. Additionally, a long-term hovering strategy is proposed, maximizing hovering time while minimizing fuel usage by exploiting naturally periodic relative orbits and a fast target-point selection method. The research is then extended to direction-constrained scenarios where thrust is limited to predefined directions, for example, tangential, radial, or normal. Energy-optimal solutions in these configurations are derived analytically, while fuel-optimal solutions are obtained through a semi-analytical Non-Linear Programming (NLP) process, initialized by the energy-optimal result. In particular, the tangential case benefits from an asymptotic analytical propagation technique, enabling fast solution generation. These predefined direction solutions are further used to formulate a two-phase control strategy to achieve 6-degree-of-freedom orbital control. The thrust is constrained to a single predefined direction during each phase, one dedicated to in-plane motion control and the other to out-of-plane motion control. In addition, a user-defined intermediate coast phase is incorporated between the thrust phases, enabling reorientation of the thrust direction and accommodating supplementary mission constraints. The study is further expanded to address control problems involving constant thrust directions defined in both the local radial–transverse–normal frame and the Earth-centered inertial frame. A least-square method is employed to identify candidate thrust directions based on energy-optimal solutions, which subsequently serve as initial guesses for the NLP procedure to obtain fuel-optimal solutions. Several state-propagation schemes are developed and compared to assess their trade-offs in terms of computational cost and solution fidelity. An efficient overall strategy is proposed to determine the optimal constant control direction along with its corresponding optimal bang-off-bang thrust profile. The effectiveness and robustness of proposed algorithms are validated through comprehensive numerical simulations and Monte Carlo analyses. The results demonstrate that the developed methods provide accurate, computationally efficient, and easily implementable solutions for proximity operations involving direction-constrained low-thrust control. These features make the approaches particularly well-suited for onboard implementation in CubeSat missions and small spacecraft requiring autonomous guidance with limited actuation authority.| File | Dimensione | Formato | |
|---|---|---|---|
|
PhD_Thesis_ChunchengZhao.pdf
accessibile in internet solo dagli utenti autorizzati
Descrizione: PhD Thesis Main Body
Dimensione
10.4 MB
Formato
Adobe PDF
|
10.4 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/245177