Demonstration of Chromosome Representations of Genetic Algorithms for Solving Mathematical Models

This paper was aimed to demonstrate the modification on chromosome representations for solving two non-linear continuous mathematical models with single and multiple variables using binary chromosome genetic algorithms (GA). The binary chromosome was used to encode single variable and two variables models, whilst both optimum solutions from each model were initially identified. The experimental results obtained from factorial design after applying GA to solve both models, each of which with five replications, were analyzed using a general linear form of analysis of variance and main effect plots. It was found that the appropriate setting of GA parameters was case dependent due to the nature of the problems and the size of its solution space. It was also found that the random seed, which is not GA parameter but is a nuisance factor occurred during the random procedure, affected on the performance of the algorithms.

Keywords: Genetic algorithms, Design and analysis of experiment, Optimization

Corresponding author: E-mail: pupong@nu.ac.th