An Improved Genetic Algorithm for Optimization of Mathematical Test Functions

EasyChair Preprint 504

Abstract

In this paper, we present an Improved Genetic Algorithm (IGA) for solving the problem of sub-optimal convergence as well as over fitting/elitism of the parent selection method. This entails the development of a K-means clustering selection method where chromosomes are clustered into two non overlapping groups with the best group being selected for the reproduction process. This work is geared towards an on-going effort in developing a Vehicle Ad-hoc Network (VANET) route optimization algorithm for road anomaly monitoring. Towards the realization of this goal, the developed improved GA was tested on mathematical test functions as part of the preliminary performance evaluation of the algorithm as reported here. Specifically, it was observed that the IGA converges to a better average solution after 40 iterations when compared to that of the conventional GA with roulette wheel selection technique. Thus, suggesting an improved performance when applied for road anomaly route optimization in a VANET system.

Keyphrases: Genetic Algorithm, Optimization, Solution and, Test functions, route

@booklet{EasyChair:504,
  author    = {Habeeb Bello-Salau and Abiodun Musa Aibinu and Elizabeth Nonye Onwuka and Jaiye Jaheosphat Dukiya and Adeiza James Onumanyi},
  title     = {An Improved Genetic Algorithm for Optimization of Mathematical Test Functions},
  howpublished = {EasyChair Preprint 504},
  year      = {EasyChair, 2018}}