Dottorato in INGEGNERIA AERONAUTICA E SPAZIALE

Titolo della tesi: Control-oriented modelling of multibody space systems: theory and experiments on attitude and vibration control of flexible spacecraft

A significant push both in terms of development of advanced space-qualified technologies and in terms of modelling approaches and control design is a current necessity to cope with the pointing requirements demanded for future space missions. Naturally such developments are particularly hard in space systems due to the cost, complexity and safety-critical nature of such systems. As far as control design solutions and Validation and Verification (V&V) approaches are concerned, space system manufacturers are historically extremely conservative and the use of simple classical control techniques is by far the most common state-of-practice as they are vastly established and understood. However, these industrial practices offer limited capabilities and often sub-optimal solutions when addressing complex Multiple-Input Multiple-Output (MIMO) models, uncertain systems and tight design requirements to be met. Robust control techniques provide the mathematical framework to address the aforementioned issues. Indeed, an uncertain description is straightforwardly given via the Linear Fractional Transformation (LFT) operation; multiple requirements and objectives are systematically defined via frequency domain specification in the form of weighting functions and formal guarantees of stability/performance in spite of the uncertainties are provided by the structured singular value. This dissertation provides an effort to address both the numerical and the technological aspects. From the modelling and control point of view, the LFT framework and recent linear robust control techniques are adopted for the design and control of flexible multibody systems. Regarding the technological aspect, this researcher investigates the use of smart materials as Active Vibration Control (AVC) devices, in particular piezoelectric materials, which are becoming a more and more interesting and mature technology for various space applications. In this frame an experimental test-bed, replicating the dynamics of a LSS, will be also employed to test and verify all the theoretical and technological aspects developed in the present thesis. In the thesis the LFT framework, with the use of the Satellite Dynamic Toolbox (SDTlib) developed at ISAE-SUPAERO, is employed to model multi-body systems with substructures connected in closed-loop kinematic configurations. The derivation of a LFT/LPV model for a 6-DoFs tri-dimensional parallel robots, the Stewart-Gough platform, is performed. The rational behind the approach used to solve the loop closure equations and provide the parametrized model according to the end-effector pose of the platform is described by the author. Eventually, a second macro area is researched in this thesis, which involves the monitoring of the health status of a spacecraft. Firstly, the implementation of a Fault Detection and Isolation (FDI) technique is investigated for identifying possible failure in piezoelectric actuators. Indeed, the addition of the AVC system proposed in the first part of the dissertation introduces additional complexity and further points of failures for a space system which should naturally be addressed. A second topic stems from the increasingly growth in size and flexibility of space structures and concerns a preliminary investigation of machine learning-based strategies for Structural Health Monitoring (SHM) of these large structures.