Performance comparison of stochastic search algorithms on the interplanetary gravity-assist trajectory problem

The comparison between two evolutionary algorithms (EAs) differential evolution (DE) and particle swarm optimization (PSO) on an earth to Jupiter gravity assist trajectory optimization problem are discussed. Evolutionary algorithms are used to optimize in terms of a trajectory between earth and Jupiter which uses a single gravity assist from either earth or Venus. Several assumptions were made in the construction of this problem that includes the use of two body dynamics and optimal gravitational flybys. DE and PSO should be run ten times each because of the computationally intensive nature of this problem for the earth swing-by and Venus swing-by cases. DE and PSO proved robust given their reliability record in the random seed analysis. It was observed that PSO would perform the best on the interplanetary problem because it outperformed DE on Rosenbrock's saddle, Rastrigin's function, and the LEO-LEO transfer.