Metaheuristics to Design Satellite Constellations
Résumé
At the beginning of the space conquest the space-systems were composed of unique satellite. With the growth of the needs and the explosion of the technologies, the space-systems became more and more complex. Now they are composed of many satellites working together : we talk about constellation. The aim of the constellation design, is to optimize the satellite positions in order to obtain the best performances at the lowest cost. In fact, launching a satellite is expensive and the reduction of the constellation size is necessary. The field of satellite constellation optimization is complex because of the size of the exploration space, the presence of local optimum, and the time consuming criterion. For this reason, applying any optimization algorithm over the whole constellation using a black-box criterion leads to an impossible resolution. In order to bypass these obstacles and to simplify the search we decide to split the constellation and the exploration space. We also use approximated criterions to reduce the evaluation of a solution and to analyze its behavior while using a meta-heuristic. The article first presents the terminology and the modelling step of the problem. Secondly we talk about resolution. The last part shows an overview of the results. As a conclusion we present the perspectives of the algorithm.