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.

2.3.2 Branch-<strong>and</strong>-Bound<br />

The branch-<strong>and</strong>-bound <strong>method</strong>ology considered here is a variation on the general<br />

branch-<strong>and</strong>-bound <strong>method</strong> that uses linear programming relaxations, as presented in<br />

Section 1.2.3. In an integer programming problem the relaxation step involves removing<br />

the integer constraints <strong>and</strong> solving <strong>for</strong> the solution to the continuous problem. Relaxing<br />

the integer constraints is not possible, however, when the integer variables represent<br />

discrete asteroids choices. There<strong>for</strong>e, in place of the linear programming relaxation, a<br />

two-impulse Lambert solution is used to provide an upper bound on the low-thrust<br />

solution <strong>for</strong> the branch-<strong>and</strong>-bound algorithm. The proposed branch-<strong>and</strong>-bound algorithm<br />

is based on the conjecture that the two-impulse solutions provide a reliable upper-bound<br />

to the low-thrust problem. This assumption will be examined on the sample problem.<br />

S<br />

ast 1 = 1 ast 1 = 2 ast 1 = 3<br />

S 1 S 2 S 3<br />

ast 2 = 4 ast 2 = 5 ast 2 = 6 ast 2 = 4 ast 2 = 5 ast 2 = 6 ast 2 = 4 ast 2 = 5 ast 2 = 6<br />

S 14 S 15 S 16 S 24 S 25 S 26 S 34 S 35 S 36<br />

ast 3 = 7 ast 3 = 7 ast 3 = 7 ast 3 = 7 ast 3 = 7 ast 3 = 7 ast 3 = 7 ast 3 = 7 ast 3 = 7<br />

ast 3 = 8 ast 3 = 8 ast 3 = 8 ast 3 = 8 ast 3 = 8 ast 3 = 8 ast 3 = 8 ast 3 = 8 ast 3 = 8<br />

S 147 S 148 S 157 S 158 S 167 S 168 S 247 S 248 S 257 S 258 S 267 S 268 S 347 S 348 S 357 S 358 S 367 S 368<br />

Figure 13: Example branch-<strong>and</strong>-bound tree.<br />

The search tree enumerates all possible asteroids sequences, a small example of<br />

which is illustrated in Figure 13. The first branch represents the choice of the first<br />

asteroid to visit from Earth. The next branch represents the second asteroid to visit in the<br />

sequence <strong>and</strong> so on. The branch-<strong>and</strong>-bound tree is used only to solve <strong>for</strong> the optimal<br />

asteroid sequence – the optimal departure date, flight times, <strong>and</strong> stay times must be<br />

obtained using another <strong>method</strong>. In order to begin the algorithm, a known low-thrust<br />

48

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

Saved successfully!

Ooh no, something went wrong!