Constellation Optimization using Genetic Algorithm : Combining SAR & Optical Sensors with AI Requirements

Abstract: With increasing world tensions and improvements of satellites and their sensors, the interest and possibility of using space and satellites for defensive purposes has increased greatly. However, not much research has been conducted into the needs and possibilities of satellite constellations over Sweden, especially using SAR and optical sensors combined with AI object detection. This thesis provides insight in to the needs and requirements to achieve certain coverage and gap times and explores different constellation design methods to do so. This is done by combining large scale tests performed with genetic algorithm and a dual-axis propagator with theoretical and analytical methods. Results show that for micro-satellites under 100 kg based on current commercial technology, it is found that between 24 to 63 satellites are needed for 1 hour gap times depending on what combination of SAR and optical satellites are used. The genetic algorithm was found to not generate optimal constellations as the number of satellites increased beyond 12. It was however useful in mapping out possibilities and finding certain optimal parameters such as the inclination. The dual-axis propagator tested for its low processing load was found to be good for coverage analysis and estimating the shapes of the orbits. It was noted to have large positional errors however, limiting its use to analysis and not full constellation design.