Abstract
K-means clustering combined with genetic algorithm (GA) techniques are used to improve the accuracy of estimation process and to minimize computational effort for solving nonlinear optimization problems. The main purpose of K-means clustering is to exhibit faster convergence which turns into quick evolution. This paper focuses on newly proposed cluster based GA selection techniques for solving unconstrained optimization problems. The K-means cluster based genetic algorithm (GKA) selection techniques comprise of four major stages: clustering, membership probability indexing, fitness evaluation and selection. The hybridization of genetic algorithm and clustering will effectively cater the problem of population diversity and selection pressure. There are two types of GKA selection techniques that are examined, the first selection technique (GKAF) includes two proposed selection operators which are linked with a fixed number of clusters while the second technique (GKA0) is based on the optimum number of clusters. The main focus of these new selection techniques is to preserve population diversity as well as to avoid local optima. The performance of each technique is evaluated through eleven well known benchmark functions. On the whole, the novel cluster based selection techniques are demonstrated to be extremely efficient and effective for achieving optimum solutions which are verified by simulated results.