基于改进蝙蝠优化自确定的模糊C-均值聚类算法

Self-Defined Fuzzy Clustering C-Means Algorithm Based on Improved Bat Optimization

针对模糊C-均值聚类算法敏感于初始聚类中心及聚类收敛慢、聚类数目手动设定等缺陷,提出了基于改进蝙蝠优化自确定的模糊C-均值聚类算法。该算法是基于密度峰值综合衡量聚类中心外围数据密集程度和聚类中心间距离,自动确定聚类中心和聚类数目,以此作为改进蝙蝠算法的初始中心;在原始蝙蝠算法中引入Levy飞行特征加强算法跳出局部最优能力;使用Powell局部搜索加快算法的收敛,利用改进的蝙蝠种群进行种群寻优,并将最优蝙蝠位置作为聚类C-均值新聚类中心,进行模糊聚类,以此循环交叉迭代多次最终获得聚类结果。将基于改进蝙蝠优化自确定的模糊C-均值聚类算法与其它两种聚类算法在标准数据集上进行仿真对比,实验结果表明:与其它两种算法相比,该算法收敛速度快、误差率低。

For the fuzzy clustering C-means algorithm is sensitive to the initial clustering center and clustering slow convergence,manually set the number of clusters and other defects,self-defined fuzzy clustering C-means algorithm based on improved bat optimization was proposed.Based on the density peak value,the density of data and the distance between cluster centers were measured,so as to automatically determine the number of cluster centers and clusters,which was used as the initial center of the improved bat algorithm.The Levy flight characteristics were introduced to enhance the bat algorithm to jump out of the local optimum ability,and Powell local search was used to accelerate bat algorithm convergence.The improved bat population was used for population optimization,and the optimal bat position was used as the clustering C-means new clustering center,and fuzzy clustering was carried out to obtain the clustering results by repeated iterative iterations.Compared with the other two clustering algorithms on the standard dataset,the experimental results showed that the proposed clustering algorithm can converge quickly with lower error rate.