基于模糊最近邻聚类学习算法的海水藻类繁殖状态预测研究.pdfVIP

第 卷 增刊 东南大学学报( 自然科学版) Vol. Sup 2011 年 月 JOURNAL OF SOUTHEAST UNIVERSITY (Natural Science Edition) 2011 基于模糊最近邻聚类学习算法的海水藻类繁殖状 态预测研究 张颖1 李鹏2 邬益川2 (1 上海海事大学信息工程学院, 上海 200135 ) (2 国家海洋局东海标准计量中心, 上海 201306) 摘要: 为了预报海洋赤潮等灾害的发生，需要监测海水中藻类的繁殖状况。本文通过建立针对海 水中叶绿素-a 含量的预测模型间接预测海水中藻类的繁殖生长状况。藻类生长受海水中多种环境 理化因子的作用和影响，且各种理化因子之间也存在相互关联影响，这类复杂非线性系统很难用 机理模型去描述。本文运用基于数据的模糊最近邻聚类学习算法进行海水叶绿素-a 含量状态的预 测模型构建，并针对采样数据的特点和降低模型阶次的实际需求，设计了改进的模糊最近邻聚类 学习算法。实验表明这种方法可以降低模型构造的复杂度，且使模型具有较好的泛化性能，能够 有效预测海水中藻类的繁殖生长状况. 关键词: 叶绿素-a; 模糊最近邻聚类; 状态预测; 泛化性能 中图分类号: X55 文献标识码: A Research of state prediction of algae reproduction in seawater based on learning algorithm of fuzzy nearest neighbor clustering Zhang Ying1 Li Peng2 Wu Yichuan2 1 ( College of Information Engineering, Shanghai Maritime University, Shanghai 200135, China) 2 ( Standard Measurement Center of the East China Sea of the State Oceanic Administration, Shanghai 201306, China) Abstract: In order to forecast the occurrence of red tide in the ocean, the state of algae reproduction in seawater need to be observed. The method of monitoring the state of algae reproduction by constructing the predictive model to predict the content of chlorophyll-a was proposed. Algae growth is effected by many environmental physical-chemical factors, and there exists interrelationship among these factors, this kind of complex non-linear system is difficult to represent by mechanism model. Fuzzy nearest neighbor clustering method based on data driven was used to constru