An Incremental Algorithm for Fast Optimisation of Multiple Gravity Assist Trajectories
- Paper number
IAC-07-C1.3.04
- Author
Mr. Matteo Ceriotti, University of Glasgow, United Kingdom
- Coauthor
Dr. Massimiliano Vasile, University of Glasgow, United Kingdom
- Coauthor
Mr. Claudio Bombardelli, University of Padova, Italy
- Year
2007
- Abstract
Multiple gravity assist (MGA) trajectories are and have been essential to reach high-Δv targets with low propellant consumption. In mathematical terms, the problem of finding a good first guess solution for the design of a MGA trajectory can be seen as a global optimisation problem. The dimension of the search space, and of the possible alternative solutions, increases exponentially with the number of swing-bys. This makes the search for a globally optimal transfer quite difficult. The number of possible paths, and thus the complexity of the problem, further increases if one or more deep-space manoeuvres are inserted in between two subsequent planets. However, the whole problem can be decomposed into smaller sub-problems and solved incrementally. In fact, starting from the departure planet and flying to the first swing-by planet, only a limited set of transfers are feasible, for example with respect to the maximum achievable excess velocity. Therefore, when a second leg is added to the trajectory, only the feasible set for the first leg is considered and the search space is reduced. The process iterates by adding one leg at a time and pruning the unfeasible portion of the solution space. Previous works have employed a grid search in order to identify the unfeasible regions of the search space for each sub-problem. This approach is not sufficiently efficient if deep-space manoeuvres are considered and each leg is tightly coupled with the others. In fact, in this case, either a very fine grid is required to identify the unfeasible parts of the domain or a mild pruning criterion has to be adopted in order to preserve the good solutions. As a result, the computational time quickly becomes unacceptable even for a limited number of planets. In this work, the grid sampling is substituted with a global search through a stochastic population-based optimiser. At each level, in the incremental process, the search space in the neighbourhood of the local optima is preserved and considered for the next level, while the remaining part is pruned out. The algorithm has been applied to a number of test cases to investigate the efficiency of the exploration of each sub-problem, and the reliability of the space pruning. A comparison to the direct global optimisation of the whole trajectory is shown for some realistic mission design problems.
- Abstract document
- Manuscript document
IAC-07-C1.3.04.pdf (🔒 authorized access only).
To get the manuscript, please contact IAF Secretariat.