中文摘要：

提出费用函数的概念，构建采用费用函数改进FCM（Fuzzy C—Median）的优化算法，并且指出最伟聚类数的上限和下限，大大缩小了最佳聚类数的搜索空间，同时，在理论上证明了经验规则的合理性．把改进FCM算法应用到实际生活中，通过对聚类结果的仿真研究进一步验证了模型的有效性和实用性．

英文摘要：

The expense function was brought forward, which was used to improve the classical FCM （Fuzzy C-Median） algorithm, and upper limit and lower limit were pointed out, which reduced the searching space hy a long way. At the same time, the experience rule was proved rational in theory. The improved FCM algorithm was applied in practice, and the simulation of cluster results proved the validity and practicability of the improved FCM.