Résumé (EN)

The object of this paper is to present a multidimensional, global approach to the optimization of trajectories for Earth observation satellites. In the known approaches for the determination of these trajectories, values of constraints are fixed for some characteristic criteria and the optimal trajectories are then obtained in the technical sense. In this approach, however, there are computed one or several trajectories by interactive variation of the values of these criteria. The set of trajectories obtained by this manner are the best with respect to all criteria and constitute an optimum in the case of a single decision maker or a satisfactory compromise in the case of multiple decision makers. The present paper first considers the basic elements concerned in the modelling of the problem and then, in the second part, presents a global optimization method in which the interactive algorithm of Bernabeu and Winkels for the evolutive reduction, improvement and classification is combined with the ordinal optimization approach of Marcotorchino and Michaud.