Evolutionary Algorithm And Its Application In Clustering Problem

This thesis includes four chapters, and the main content focuses on the improvement of the model of evolutionary algorithm and the enhancement of the cooperation mechanism to coordinate local search and global search. In addition, we combine Particle Swarm Optimization (PSO) with Fuzzy C-Means (FCM) clustering algorithm, and employ the statistical test method for single modal based on T-square sample theory to adaptively analyze the unknown data set.In the second chapter, a coevolutionary algorithm with memory (MCEA) is proposed. We divide the population into a subgroup and other individuals so as to local search and global search respectively, and design cooperation operator and mutation operator for efficient and effective crossover and combination in the population. Numerical experiment illuminates the validity of MCEA according to the comparison of the performance between with Evolutionary Programming Made Faster (FEP) and Organizational Evolutionary Algorithm (OEA).In the third chapter, for the clustering problem as one of the most fundamental problems in Pattern Recognition, we put forward an Adaptive Fuzzy C-Means Clustering Algorithm Based on Particle Swarm Optimization (AFCM-PSO). The basic knowledge of FCM and PSO is introduced in this chapter, and such two algorithms are tried to be combined in AFCM-PSO. The statistical test method for single modal based on T-square sample theory is employed to analyze the clustering tendency and validity of the unknown data set. The rationality and the validity of AFCM-PSO are showed in numerical experiment.Finally, the work of this paper is summarized, and the prospect of the relevant research is given.