The University of Edinburgh -
Division of Informatics
Forrest Hill & 80 South Bridge

MSc Thesis #9562

Title:A Genetic Algorithm for Optimising Inter-Planetary Trajectories
Date: 1995
Abstract:The design of an interplanetary trajectory for a space probe, such as that for the Galileo mission to Jupiter, is a complicated and time consuming process. Initial work by Wilson ([Wilson 93]) suggested that the group of computational models known as genetic algorithms could be applied to this type of problem with some success. It was the aim of this project to implement a genetic algorithm to evolve the set of course corrections necessary to pilot a simulated space probe through a 2-D solar system. The solution was not only required to take the probe to a specific destination planet but also to do so with an efficient use of fuel. Several variations on the basic genetic algorithm model were tested in this project with a combination of steady state control strategy and one point crossover being found to be the most effective. Promising results were obtained from this variation on a problem of similar complexity to the Galileo mission. Using three small sub-populations of 25 chromosomes the algorithm is able to evolve an effective trajectory in only 30 generations. Using this trajectory the probe not only reached its target but also appears to do so in a relatively efficient manner. Although limited in certain ways, the results from this project were enough to give good encouragement that genetic algorithms do have good potential for this type of trajectory optimisation problem.

[Search These Pages] [DAI Home Page] [Comment]