Minimum Time Spacecraft Reorientation with Hybrid Heuristic/Gradient Optimization Methods
Open Access
- Author:
- Guyer, Luke
- Area of Honors:
- Aerospace Engineering
- Degree:
- Bachelor of Science
- Document Type:
- Thesis
- Thesis Supervisors:
- Robert G. Melton, Thesis Supervisor
Mark David Maughmer, Thesis Honors Advisor - Keywords:
- Metaheuristics
Optimization
Particle Swarm Optimization
Ant Colony Optimization
Spacecraft Dynamics
Hybrid Optimization
Spacecraft Reorientation - Abstract:
- Spacecraft often need to be reoriented in a time-optimal manner to satisfy mission requirements. By formulating a maneuver as a mathematical optimization problem, reorientation has been studied extensively in the past in search of fast yet sufficiently optimal solutions. Attitude constraints may also be necessary to avoid the damaging of optical sensors by bright celestial bodies. This research considers a hybrid heuristic/gradient optimization algorithm applied to an inverse dynamics formulation of spacecraft reorientation as a fast approach at finding a sufficiently optimal reorientation time in constrained space. Two heuristic algorithms, Particle Swarm Optimization (PSO) and the Continuous Interacting Ant Colony (CIAC) algorithm, were paired with a gradient-based optimizer to make up two different versions of the hybrid solver. For the constrained case in which 500 or more heuristic iterations were used, the algorithm found a feasible solution 90% of the time and a quasi-optimal solution 82% of the time, requiring 1.5 minutes of runtime in MATLAB. The gradient-based optimizer provided a significant improvement to a purely heuristic solution, decreasing a CIAC maneuver time by 40% and a PSO maneuver time by as much as 60%. For the case of unconstrained reorientation, the optimizer exhibited out-of-plane motion and control torques approaching bang-bang structures. The CIAC version of the hybrid algorithm consistently performed faster and found lower average maneuver times than PSO in less heuristic iterations.