Abstract: In order to overcome premature convergence of genetic algorithm and derive better performance and higher accuracy on clustering problems,a new clustering method based on immune genetic algorithm is proposed in this paper.Immune principle is introduced into genetic clustering algorithm in the method,through adding density regulatory factor on selection operator performed according to the fitness in the mechanism.Selection probability is adjusted,and the diversity of individuals is maintained,and then the disadvantage of premature convergence is overcome.The algorithm represents individuals by improved grouping coding;based on it,a corresponding manner of initial population is formulated and appropriate genetic operators are designed,which enable the algorithm automatically finding the proper number of clusters and the proper partition from a given data set.The theoretical analysis and experiment result show that the algorithm derives better performance and higher accuracy on clustering problems.

Key words: immune principle, genetic algorithm, clustering method