20.04.2014 Views

design space pruning heuristics and global optimization method for ...

design space pruning heuristics and global optimization method for ...

design space pruning heuristics and global optimization method for ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

these two problems, the best known solutions were all located within the top 2% of the<br />

ranked sequences in the branch-<strong>and</strong>-bound algorithm. Since the 7 th <strong>and</strong> 12 th best known<br />

solutions (listed in Table 26) fall outside this range, the number of low-thrust<br />

<strong>optimization</strong>s required to locate these two solutions is unknown.<br />

Figure 57 plots the number of asteroid sequences that require low-thrust<br />

<strong>optimization</strong> as a function of the number of asteroid sequences evaluated in the branch<strong>and</strong>-bound,<br />

<strong>for</strong> the first 78,000 ranked sequences (plotted in red are the data points from<br />

Table 26). In other words, Figure 57 illustrates the efficiency of the branch-<strong>and</strong>-bound<br />

algorithm at <strong>pruning</strong> out branches of the tree using the relaxed two-impulse solutions. In<br />

the range considered, approximately seven percent of the sequences in the branch-<strong>and</strong>bound<br />

tree require low-thrust <strong>optimization</strong> – the rest are pruned out due to their optimal<br />

two-impulse solution being less than the lower bound of 100 kg/yr. As the number of<br />

sequences evaluated in the branch-<strong>and</strong>-bound tree increases, the percent that require lowthrust<br />

<strong>optimization</strong> decreases, due to the fact that the optimal two-impulse solutions tend<br />

to decrease as a function of their ranking in the branch-<strong>and</strong>-bound algorithm. This trend<br />

was observed previously in Figure 53, which plots the evolution of the branch-<strong>and</strong>-bound<br />

algorithm <strong>for</strong> the modified GTOC2 problem. The number of low-thrust <strong>optimization</strong>s<br />

required as a function of the number of branch-<strong>and</strong>-bound sequences evaluated can also<br />

be plotted <strong>for</strong> the modified GTOC2 problem, in order to illustrate this trend <strong>for</strong> a problem<br />

where the entire branch-<strong>and</strong>-bound tree was evaluated. As expected, Figure 58 illustrates<br />

how the percent of sequences that require low-thrust <strong>optimization</strong> decreases towards the<br />

end of the branch-<strong>and</strong>-bound algorithm.<br />

132

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!