人工免疫算法在洪水分类中的应用

 在总结洪水分类研究的基础上，提出了一种新的洪水分类方法，以人工免疫网络对洪水样本进行免疫学习和记忆，提取表征洪水强度的有用特征得到抗体库和相似度矩阵，利用最小生成树方法，依据抗原与记忆集的亲和度确定洪水的分类。以宜昌站12场典型洪水过程和广东石狗站17场典型洪水过程为例进行了洪水聚类分析，结果表明：所提算法有效提取了同类型洪水的模糊特征和规律，去除了不必要的信息冗余，较好地将同类洪水聚集在了一起；与进化粒子群优化算法相比，该法有更快的收敛速度。

Application of Artificial Immune Network in Flood Classification

After summarizing flood classification research,a novel flood classifier employing artificial Immune network was proposed to carry out immune memory and learning of the flood sample. The useful characteristics that effectively represent the flood intensity are extracted to produce antibody repertoires and similarity matrix. Then minimum spanning tree is used to classify flood sample by calculating the affinity between antigen and memory cell sets. The proposed classifier was testified by 12 typical hydrographs of Yichang station and 17 typical hydrographs of Shigou station. The results show that the proposed classifier classified the similar flood into same class by extracting the fuzzy characteristics and removing redundant information. Comparing with Evolutionary Particle Swam Optimization, the classifier fastens the convergence rate.