一种改进的FCM聚类算法及其在赤潮预测中的应用

针对传统FCM算法的不足，文中提出了一种改进的FCM算法，利用样本与聚类中心相似关系确定各样本对聚类中心的影响系数并对欧氏距离进行加权，使得聚类过程更加迅速，聚类结果更加稳定精确.最后的数据仿真结果也证明本文所作的改进是行之有效的.将改进的算法应用到赤潮预测预警中，直观地反映了海洋要素样本的分布情况，为赤潮预测预警展示了一种新方法.

Improved FCM clustering algorithm and its application in red tide prediction

The distance definition in Euclidean space was weighted and the weight coefficient of samples effecting the fuzzy clustering centers was estimated by using the similarity relations between samples and fuzzy clustering centers. As a result,the improved method was proved to have a better result by using the method of data simulation.Meanwhile,the improved method was introduced into red-tide prediction,which visually reflects the distribution of samples of marine elements.