Evolutionary Computation for Spacecraft Trajectory Optimization

Open Access
Author:
Sottile, Bradley Joseph
Graduate Program:
Aerospace Engineering
Degree:
Master of Science
Document Type:
Master Thesis
Date of Defense:
None
Committee Members:
  • Robert Graham Melton, Thesis Advisor
Keywords:
  • Astrodynamics
  • Evolutionary Algorithms
  • Evolutionary Computation
  • Evolutionary Strategies
  • Optimization
  • Spacecraft Trajectory
Abstract:
Evolutionary Computation has exploded in use in engineering and the applied sciences. For this thesis, three algorithms – Particle Swarm Optimization (PSO), Bacteria Foraging Optimization (BFO) and Covariance Matrix Adaptation Evolution Strategy (CMA-ES) – are compared against each other to solve a classic problem in astrodynamics, the Hohmann transfer. The role of fixed and varying penalties is explored for each algorithm and compared. Each algorithm was run 1000 times and the performance metrics were compared. PSO using fixed penalties ran with an average central processing unit (CPU) time of 0.138 seconds and yielded a mean error of 1.30% and a median error of 0.48%. Using varying penalties, the algorithm ran with an average CPU time of 0.107 seconds and yielded a mean error of 1.78% and a median error of 0.43%. BFO with fixed penalties had a mean CPU time of 0.655 seconds and yielded a 2.19% mean percent error and 1.91% median percent error. For the varying penalty case, BFO averaged a CPU time of 0.727 seconds, a mean percent error of 0.27% and a median 0.36%. CMA-ES with fixed penalties yielded a mean CPU time of 0.572 seconds, a mean percent error of 0.26% and a median percent error of 0.00%. The varying penalty case for CMA-ES yielded a mean CPU time of 0.582 seconds, a mean percent error of 0.43% and a median percent error of 0.43%. The algorithms all excelled in some areas and had poor performance in others, especially as the penalty case varied. A clear result is that algorithm selection is problem-dependent. Suggestions for future work and applications to other problems are provided.