27.06.2013 Views

Evolution and Optimum Seeking

Evolution and Optimum Seeking

Evolution and Optimum Seeking

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.

Chapter 5<br />

<strong>Evolution</strong> Strategies for Numerical<br />

Optimization<br />

The task of mimicking biological structures <strong>and</strong> processes with the object of solving<br />

technical problems is as old as engineering itself. Mimicry itself, as a natural \strategy",<br />

is even older than mankind. The legend of Daedalus <strong>and</strong> Icarus bears early witness<br />

to such human endeavor. A sign of its scienti c coming of age is the formation of the<br />

distinct branch of science known as bionics (e.g., Hertel, 1963 Gerardin, 1968 Beier<br />

<strong>and</strong> Gla , 1968 Nachtigall, 1971 Heynert, 1972 Zerbst, 1987), which is concerned with<br />

the recognition of existing biological solutions to problems that also happen to arise<br />

in engineering, <strong>and</strong> with the adequate emulation of these examples. It is always thereby<br />

supposed that evolution has found particularly good, perhaps even optimal solutions. This<br />

assumption has often proved to be correct, or at any rate useful. Only a few attempts to<br />

imitate the actual methods of natural development are known (Ashby, 1960 Bremermann,<br />

1962{1973 Rechenberg, 1964, 1973 Fogel, Owens, <strong>and</strong> Walsh, 1965, 1966a,b Holl<strong>and</strong>,<br />

1975 see also Chap. 4) since they are curiously regarded a priori as being especially bad,<br />

meaning costly.<br />

Rechenberg proposed the hypothesis \that the method of organic evolution represents<br />

an optimal strategy for the adaptation of living things to their environment," <strong>and</strong> he<br />

concludes \it should therefore be worthwhile to take over the principles of biological<br />

evolution for the optimization of technical systems."<br />

5.1 The Two Membered <strong>Evolution</strong> Strategy<br />

Rechenberg's two membered evolution scheme, suggested in similar form by other authors<br />

as a r<strong>and</strong>om strategy (see Chap. 4) will be expressed in this chapter as an algorithm for<br />

solving non-discrete, non-stochastic, parameter optimization problems. As in Chapter 3,<br />

the problem is<br />

F (x) ! min<br />

where x 2 IR n . In the constrained case x has to be in an allowed region G IR n , where<br />

G = fx 2 IR n j Gj(x) 0 j = 1(1)nGj restriction functionsg<br />

105

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

Saved successfully!

Ooh no, something went wrong!