Methods and tools for space systems optimization, with applications

Research focus

The Peer review has evaluated this group as Average

The research interests in this theme are in the following focus areas: Interplanetary trajectories and gravitational maneuvers. Several aspects of optimal interplanetary trajectory design have been addressed, based on local optimization and global optimization concepts. The Department has developed a software tool under ESA contract to design optimal low-thrust gravity-assisted trajectories, based on direct finite elements in time transcription of the resulting optimal control problem. Among others, a mission to the Sun and a mission to Mercury were studied. For both the aim was to minimise the propellant consumption to reach a given target. In the first case the target was a resonant heliocentric orbit while in the second case the target was the sphere of influence of Mercury. Furthermore, global optimization algorithms based on heuristic theories have been studied, to design optimal interplanetary trajectories. Global optimizers are exploited to cope with mixed variable domains that include the type and numbers of planets to be flown by. Concurrent Engineering (CE). The research aims at defining autonomous spacecraft design methods. The attention is on two parallel areas concerning CE system design: - implementation of tools to support and address either the Team Leader or the single designer in taking decisions throughout the process; - simulation and optimization of the design process either to reduce the alternative paths at the beginning of a very complex system design or to substitute the team work. The research is based on Decision Making techniques, supported by Artificial Neural Network and Genetic Algorithms to build inference motors that drive the choices. Because of the multidisciplinary nature of the system design problem, optimization in such area benefits of multiobjective/multidisciplinary optimization techniques. The mixed nature of the design variables asks for algorithms that don’t need gradient information to perform the search. To evaluate different algorithms performances both Evolutionary Algorithms and Particle Swarm Optimization have been successfully applied, with distributed architecture to lighten the computational effort. Distribution is managed thanks to Game Theory different protocols, depending on the interaction to be simulated (collaborative, competitive, leader follower, etc). Meta-models are used: different techniques are exploited such as the Adaptive Resonant Theory linked to Bayesian nets, or the Response Surface methodology. Uncertainties are directly considered while optimizing thanks to different techniques, from interval analysis up to Taylor series. The obtained optimal solutions, therefore, are robust to the considered uncertainties and no further heavy Monte Carlo run is needed. The dynamic system theory is also exploited to simulate and address the team members’ behaviours during the design process. Distributed robust multidisciplinary design. Great effort is devoted to efficiently solve multidisciplinary optimization problems inherent to space related applications. Results of classical methods have been improved through the use of innovative heuristics during the optimization process, and tools have been developed based on Fast Evolutionary Programming and Particle Swarm Optimization (PSO) methods, which can efficiently describe the Pareto optimal front of problems characterized by a large number of objective functions. Different protocols coming from the Game Theory area have been applied to settle problem dependent architectures. Complex space system design and space vehicle optimization during atmospheric manoeuvring (launchers, Entry-Descent-Landing vehicles, aero-capture and aero-gravity assisted capsules) have been solved. Researches are also focused on automating and optimizing the design process of those complex phases. Skills in developing simplified but accurate models of the different disciplines involved and in global multiobjective optimization have been developed. For instance, the optimization of atmospheric entry, descent and landing according to both the guidance and the configuration is obtained thanks to a distributed architecture; the interplanetary trajectory optimization in terms of gravity/aero-gravity assist manoeuvres tuning according to the fuel/shield mass minimisation is gained by a multilevel architecture. Innovative tools based on the most promising classes of methods have been studied. Great experience has been gained in the field of stochastic methods with the development of algorithms based on Genetic Algorithms, Evolutionary Programming and PSO. Promising results have been obtained in the field of deterministic methods, which are encouraging the possibility of developing new metamodel-based optimization tools. Research line topic 3.2 - Methods and tools for space systems optimization, with applications 2-41 Particular skills have been achieved in the numerical transcription of control problems using collocation methods, pseudo-spectral methods, multiple shooting, and parallel multiple shooting techniques, and in dealing with problems with large set of equality and inequality constraints. An in house optimizer based on interior point algorithm has been developed, to solve inequality-constrained problem.

Dipartimento di afferenza

Dipartimento di Ingegneria Aerospaziale

Docenti afferenti

Franco Bernelli-Zazzera (Full Professor)
Amalia Ercoli-Finzi (Full Professor)
Micelle Lavagna (Assistant Professor)