一种基于遗传算法的聚类方法及其应用

文中提出了一种基于遗传算法的自适应聚类新算法,该算法以聚类中心坐标为遗传算法种群的个体,采用改进的遗传算子和群体更新规则,利用遗传算法的高效全局搜索特性实现聚类,不仅克服了模糊C-均值算法对初始聚类中心和样本输入次序敏感等不足,而且在结合聚类有效性分析的基础上实现了聚类中心数目的自适应调整.通过以该自适应聚类算法对MQAM信号星座进行重构,提出了一种基于星座聚类的MQAM调制识别新方法.仿真表明,文中提出的聚类算法运算效率较高,结果令人满意;基于该聚类算法的MQAM信号识别方法是实际有效的.

A Genetic Algorithms Based Clustering Method and Its Application

In this paper,a novel adaptive clustering algorithm based on Genetic Algorithms(GA)is proposed.This method takes cluster centers as GA's individuals,employs modified GA operators,then performs clustering with the effective global searching ability of GA,thus not only overcoming the sensitivity to initial centers as Fuzzy C-Means algorithm does,but also determining the number of clusters adaptively via validity analysis.Using this algorithm to rebuild signal's constellation,a clustering based MQAM constellation recognition method is constructed.Experiments show that the clustering algorithm is effective and satisfactory,and the MQAM recognition method is practical at the same time.