Development of Hybrid Evolutionary Algorithms Tool Applied to Lunar Landing Mission -ESA EL3- Study Phase

The primary purpose of this thesis is to further advance the development of a modular hybrid meta-heuristic optimization tool and then apply it for the ESA European Large Logistic Lander (EL3) RCS thrusters' orientation optimization problem. This study was conducted in Thales Alenia Space (TAS), in response to the desire of the GNC group to have a tool for fastest optimization of complex functions during different project phases. Most real-world optimizations are highly nonlinear and multimodal, under various complex constraints. A study has been carried on hybrid optimization, focusing on the island model, coupled with a survey on metaheuristic algorithms used in this field. A first version of the tool implementing Differential Evolution and Evolution Strategy was already developed in TAS. During the thesis a Particle Swarm Optimization algorithm, a mechanism of mass mutation and the possibility to vary the cloning topology has been implemented. The tool has been tested on a series of benchmark functions and EL3 thrusters’ orientation to show its effectiveness and make a comparison between the use of individual algorithms and a combined use of them. ESA EL3 mission has the purpose to support human and robotic exploration by providing a multipurpose lander. The study conducted on the orientation of EL3 RCS thrusters required the development of an approximate solution method. It was then possible to apply the developed tool into the studying of 2 configurations, 4 and 8 thrusters, and different combinations of degrees of freedom for the orientation, with the aim of obtaining an acceptable torque margin in the lander control. The results on the benchmark functions demonstrate an increase in the efficiency and effectiveness of hybrid optimization in some benchmarks problems, in conjunction with the use of a suitable cloning topology, and an improvement in the exploration of the domain thanks to the presence of mass mutation. For the EL3 application problem, the tool was able to show the correlation between the degrees of freedom in the 4 thrusters’ configuration, while in the 8 thrusters’ configuration was not able to improve the solution in all the cases analyzed. To improve the tool’s results on high non-linear problems, future work can focus on algorithms’ parameters tuning and the analysis of different algorithms and topologies.