Application Of Genetic Algorithms For Robust Parameter Optimization

Parameter optimization can be achieved by many methods such as Monte-Carlo, full, and fractional factorial designs. Genetic algorithms (GA) are fairly recent in this respect but afford a novel method of parameter optimization. In GA, there is an initial pool of individuals each with its own speci...

Full description

Saved in:
Bibliographic Details
Main Author: Belavendram, N.
Format: Article
Language:English
Published: Universiti Malaysia Pahang 2010
Subjects:
Online Access:http://umpir.ump.edu.my/id/eprint/1652/1/11_M_M_Rahman_28072010_9_clean.pdf
http://umpir.ump.edu.my/id/eprint/1652/
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Parameter optimization can be achieved by many methods such as Monte-Carlo, full, and fractional factorial designs. Genetic algorithms (GA) are fairly recent in this respect but afford a novel method of parameter optimization. In GA, there is an initial pool of individuals each with its own specific phenotypic trait expressed as a ‘genetic chromosome’. Different genes enable individuals with different fitness levels to reproduce according to natural reproductive gene theory. This reproduction is established in terms of selection, crossover and mutation of reproducing genes. The resulting child generation of individuals has a better fitness level akin to natural selection, namely evolution. Populations evolve towards the fittest individuals. Such a mechanism has a parallel application in parameter optimization. Factors in a parameter design can be expressed as a genetic analogue in a pool of sub-optimal random solutions. Allowing this pool of sub-optimal solutions to evolve over several generations produces fitter generations converging to a pre-defined engineering optimum. In this paper, a genetic algorithm is used to study a seven factor non-linear equation for a Wheatstone bridge as the equation to be optimized. A comparison of the full factorial design against a GA method shows that the GA method is about 1200 times faster in finding a comparable solution.