Abstract: Due to conventional clustering algorithms which should get the true number of clusters and easily fall into local optimum, a new algorithm is proposed, called fuzzyGAC for shortly. The algorithm combines group evolution strategy with fuzzy clustering, and uses two phases（i.e., inheritance phase and redistribution phase） to generate a new clustering solution. The proposed approach is compared with fuzzy C-means algorithm, Particle Swarm Optimization（PSO） algorithm, and Differential Evolution（DE）. Moreover, the results indicate that fuzzyGAC can provide acceptable results in terms of both determining the correct number of clusters and the accurate cluster centroids.

Key words: fuzzy clustering, grouping evolution strategy, fuzzy C-means algorithm

摘要： 针对传统的聚类算法需要知道类的真实数目，以及容易陷入局部最优的缺陷，提出基于群进化策略的模糊聚类算法，简称fuzzyGAC。该算法将群进化策略与模糊聚类结合起来，通过两个阶段（继承阶段和重新分配阶段）来产生新的聚类结果。将提出的算法与模糊C均值算法、差分算法、粒子群算法进行比较，实验结果表明，就类的数目和聚类中心而言，该算法可以自适应地修正类的数目并且提供最优的聚类中心。

关键词: 模糊聚类, 群进化策略, 模糊C均值算法