On the Equars Attitude Dynamics and Control
- Paper number
IAC-04-A.4.06
- Author
Dr. Ijar M. Da Fonseca, Instituto Nacional de Pesquisas Espaciais (INPE), Brazil
- Year
2004
- Abstract
In this paper the performance of different evolutionary optimisation techniques are analysed and compared in the framework of the determination of minimum fuel - minimum time orbit manoeuvres. In particular, two different evolutionary methods, a conventional genetic algorithm and the MoPED (Multi-objective Parzen-Based Estimation of Distribution Algorithm), are combined with either a standard orbit propagation code or a new algorithm for the solution of the Lambert problem, resulting in a total of four possible optimisation approaches.
The results obtained from the four methods are critically reviewed on some benchmark cases taken from the literature in order to clearly assess the performance of the numerical algorithms and the sensitivity of the results with respect to the required tolerances and objectives. These simple cases allow for a detailed discussion on the application of evolutionary algorithms to orbit problems and the benefits of the definition of the manoeuvre by means of an inverse approach, where the search variables are the initial and final positions along the coast arc and the transfer time, rather than the applied Δ V magnitude and direction determined by solving a shooting problem. As a matter of fact, when the Lambert problem solver is employed, the resulting approach proved to be very efficient for the preliminary definition of optimal orbit manoeuvre strategies.
A second set of results is discussed as a final comparison in order to underline the features of the resulting methodologies on a more complex task, where multiple Δ V’s are necessary to complete the assigned transfer. In this latter framework, where the issue of sensitivity of final constraints with respect to the search variables becomes more delicate, the advantages of the combined inverse approach with the statistical evolutionary algorithm become more apparent. The comparisons will be discussed both in terms of numerical performance (accuracy of the results, computation time, convergence rate, etc.) and quality of the obtained results (reliability of the code in identifying known analytical results, capability of methods of identifying the Pareto front in the Δ V - transfer time plane, etc.).
- Abstract document
- Manuscript document
IAC-04-A.4.06.pdf (🔒 authorized access only).
To get the manuscript, please contact IAF Secretariat.