A Comparison of the Differential Evolution Method with Genetic Algorithms for Orbit Optimisation
- Paper ID
5181
- author
- company
European Space Agency (ESA)/ESTEC
- country
The Netherlands
- year
2006
- abstract
As optimisation methods evolve, new methods appear which are based on other methods but may outperform them for specific problems. Differential Evolution is one of these relatively new techniques, which is an adapted version of a genetic algorithm incorporating vector differences for perturbing the vector population. Originally developed by Kenneth Price and Rainer Storn, Differential Evolution managed to finish 3rd at the First International Contest on Evolutionary Computation which was held in Nagoya, may 1996. In this paper the Differential Evolution algorithm is tested against the widely used genetic algorithm for a number of different orbit optimisation problems. For both algorithms, different settings are tested and the settings leading to the best results are compared. The cases studied contain simple problems such as a direct transfer to Mars as well as complex problems such as low-thrust trajectories to Mercury and multiple gravity assist trajectories combined with deep-space manoeuvres. An attempt is made to find generic optimal parameter settings of the Differential Evolution algorithm.