Novel consensus strategies applied to spacecraft formation flight

• Autores: Fabrizio Paita
• Directores de la Tesis: Josep Joaquim Masdemont Soler (dir. tes.) , Gerardo Gómez Muntané (codir. tes.)
• Lectura: En la Universitat Politècnica de Catalunya (UPC) ( España ) en 2017
• Idioma: español
• Tribunal Calificador de la Tesis: James Douglas Biggs (presid.) , Mercè Ollé Torner (secret.) , Stéphanie Lizy Destrez (voc.)
• Enlaces
  • Tesis en acceso abierto en: TDX
• Resumen

  • In this thesis we consider the problem of developing distributed control strategies for spacecraft formation flying. In light of possible interferometric applications, the study is conducted in two different contexts: formation acquisition and keeping, and attitude synchronization. Among others, elements of Graph and Lyapunov stability theory are employed to modellize the formation dynamics and to establish analytically basic convergence properties for the proposed control laws.

    In the case of formation acquisition and keeping, the related strategy is designed to asymptotically cancel the spacecraft relative velocities, so that they may reach fixed values for their relative distances. Previously applied with success to spacecraft travelling along transfer orbits to Lagrangian points, we consider it here in two different situations. First, in order to clarify the dependence of the control's performance from the Newtonian differential acceleration, we employ the strategy in a toy model consisting of a formation orbiting around a primary and being subject only to its gravitational pull. Then, we exploit the informations obtained to design a mission objective suitable for an optimal application of our control. The latter builds upon the scientific goals of the LISA Pathfinder mission and corresponds to periodic visits of the Earth-Moon-Sun saddle point, with the control employed in a neighborhood of this point to maintain a rigid formation.

    For attitude synchronization instead, a numerical framework is developed to complement rigorous stability analyses on the associated controls. We show that, in the case of a time-invariant network topology, by employing a hierarchical graph structure for the formation it is possible to conduct computationally fast Monte Carlo simulations to describe a control's parameter space and its dependence from the formation dimension and relative initial conditions. The method is then exploited to compare the performance increase given by a novel adaptive gains design when applied to a PD-like control with constant ones.