MOPSO Technique Assessment To Cope With First Guess Generation For Interplanetary Trajectories Differently Controlled
- Paper number
IAC-07-C1.3.02
- Author
Ms. Michelle Lavagna, Politecnico di Milano, Italy
- Coauthor
Gabriele Bellei, Politecnico di Milano, Italy
- Coauthor
- Year
2007
- Abstract
The paper proposes the application of the Particle Swarm Optimization technique, extended to deal with multi-objective scenarios to cope with the complex problem of fruitfully visiting the search space related to the interplanetary trajectory design, to get a space probe to its eventual target planet while minimizing the propellant mass and cleverly exploit the celestial mechanics [1], [2]. The aim is to settle a comprehensive architecture to deal with different control philosophies with single problem formalism as far as possible. Techniques typically exploited in space trajectory control are considered: impulsive high thrust here exploited to deal with deep space maneuvers definition, continuous low thrust propulsion and the well known gravity-assist (GA) maneuvers. The optimization is focused in detecting the best control sequence to minimize the propellant mass and the transfer time, being consistent with several physical and technological constraints. Real ephemeredes are assumed for the Solar system bodies’ trajectories [3]. The search space includes the launch date; the GA sequence, in terms of number and type of planets to be visited; the set of transfer times according to the number of GAs the optimizer proposes; the deep space maneuver location in space, whenever a high thrust control is selected; the parameters related to the exponential sinusoid tuning whenever a low thrust control is preferred. It should be noted that, although very powerful in detecting a preliminary trajectory consistent with a low thrust profile, exponential sinusoids have some limitations [3],[4]. Those limitations, such as the constrained tangential thrusting direction, are here removed by feeding the preliminary solutions into different levels of the optimizer. More in details, a specific multi-layer architecture has been defined to succeed in the minimum detection of such a complex multi-variate problem. Each level is devoted to search within a subset of the variable domain both hierarchically and according to a peer-to-peer optimization processes. The Particle Swarm Optimization technique, revised to deal with multi-objective scenarios, after a critical comparison with the Evolutionary Algorithms class, well suited to deal with mixed domains and quite huge search space for multimodal and possible discontinuous functions. The comparison campaign on test functions offered by the related literature showed the definitely better behavior of the PSOs both in terms of convergence accuracy and fastness [6],[7],[8]. Results here offered show the validity of the settled architecture to deal with different strategies to control the trajectory of a interplanetary space probe. Optimal solutions - identified by the MOPSO proposed algorithm -are then locally refined just focusing on the control profile, the launch date and the time of flight. References [1] J. Kennedy and R.C. Eberhart, Particle swarm optimization, In Proceedings of the IEEE International Conference on Neural Networks, 1995. [2] J.E. Alvarez-Benitez, R.M. Everson, and J.E. Fieldsend, A MOPSO algorithm based exclusively on pareto dominance concepts. In Evolutionary Multi-Criterion Optimization. Third International Conference, volume 3410 of Lecture Notes in Computer Science, Springer, pages 459–473, Guanajuato, México, 2005. [3] R. Battin, An Introduction to the Mathematics and Methods of Astrodynamics,AIAA Education Series, New York, 1987. [4] D. Izzo, Lambert’s problem for exponential sinusoids, Journal of Guidance Control and Dynamics, 29(5):1241–1245, September-October 2006. [5] A.E. Petropoulos and J.M. Longuski. Shape-based algorithm for automated design of low-thrust, gravity-assist trajectories. Journal of Spacecraft and Rockets, 41(5), Sep.-Oct. 2004. [6] C.A. Coello Coello, A survey of constraint handling methods for the application of particle swarm optimization to constrained nonlinear optimization problems, Technical Report Lania-RI-9904, Laboratorio Nacional de Informatica Avanzada,1999 [7] K Deb, Multi-objective genetic algorithms: Problem difficulties and construction of test problems., Evolutionary Computation, 7(3):205–230, 1999. [8] E. Zitzler, K. Deb, and L. Thiele. Comparison of multiobjective evolutionary algorithms: empirical results. Evolutionary Computation, 8:173–195, 2000.
- Abstract document
- Manuscript document
IAC-07-C1.3.02.pdf (🔒 authorized access only).
To get the manuscript, please contact IAF Secretariat.