基于人工鱼群算法和模糊C-均值聚类的洪水分类

为了克服FCM算法依赖初值的缺点，将人工鱼群算法（AFS）引入模糊C-均值聚类，提出了一种新的聚类算法，应用于洪水分类研究。新算法将聚类中心看作食物源，通过样本抽样产生初始鱼群。利用人工鱼群算法全局寻优、快速收敛的特点，得到一个较优的初始聚类结果，再使用FCM算法进行局部搜索。避免了单纯的FCM算法因初值选取不当，而容易陷入局部最小的缺陷。同时新算法给出了洪水等级划分的科学依据。实验结果表明，新算法具有比FCM算法更好的性能表现，使得到的分类结果更加准确合理。

Flood Classification Based on Fuzzy C-Mean Clustering and Artificial Fish Swarm Algorithm

In order to overcome the shortcoming of the FCM algorithm which depends on starting value, artificial fish swarm algorithm has been applied to the fuzzy C-mean algorithm to propose a kind of new cluster algorithm for flood classification research. The new algorithm takes the cluster center as food source and produces the initial fish swarm through sampling. The overall optimization and rapid convergence characteristic of the artificial fish swarm algorithm obtains a superior initial cluster result. And then the FCM algorithm searches the partial again, to avoid the flaw of falling into partially smallest value, because of the improper starting value of the pure FCM algorithm. And the new algorithm has given the scientific basis of the flood rank division, The result indicated that the new algorithm has better performance compared to the FCM algorithm. The classified result is more accurate and reasonable.