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.

OptimalTwo-Impulse J (kg/yr)<br />

200<br />

180<br />

160<br />

140<br />

120<br />

100<br />

80<br />

60<br />

40<br />

2 4 6 8 10 12 14<br />

Time of Flight (years)<br />

OptimalTwo-Impulse J (kg/yr)<br />

200<br />

180<br />

160<br />

140<br />

120<br />

100<br />

80<br />

60<br />

40<br />

2 4 6 8 10 12 14<br />

Time of Flight (years)<br />

Figure 52: Optimal impulsive solutions, with <strong>and</strong> without the time of flight restriction, <strong>for</strong> the<br />

modified GTOC2 problem.<br />

Table 22 lists the settings used <strong>for</strong> the genetic algorithm as applied to this<br />

problem, <strong>and</strong> Table 23 lists the <strong>design</strong> variables, bounds, <strong>and</strong> discretization. The number<br />

of bits <strong>for</strong> each <strong>design</strong> variable was chosen so that the discretization on all of the time<br />

variables was approximately equal. For the bits chosen, the variables are discretized in<br />

approximately seven day increments. Because of the greater number of low-thrust<br />

<strong>optimization</strong>s required in this problem, the genetic algorithm is run only once <strong>for</strong> each<br />

asteroid sequence. At the end of the branch-<strong>and</strong>-bound algorithm, the genetic algorithm<br />

is then run three times on the most promising solutions, in an attempt to improve upon<br />

their objective functions. Using these settings, on average, a single run of the genetic<br />

algorithm requires 1746 function calls (end-to-end <strong>optimization</strong>s by MALTO) <strong>and</strong> 49<br />

generations to converge. With the available computing resources, each genetic algorithm<br />

run takes 70 minutes on average.<br />

123

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

Saved successfully!

Ooh no, something went wrong!