基于支持向量机分类的海州湾赤潮判别

RESEARCH ON IDENTIFICATION OF RED TIDES IN HAIZHOU GULF BASED ON SUPPORT VECTOR MACHINE CLASSIFICATION

近年来,海州湾赤潮暴发日益频繁,对当地经济发展和生态安全构成严重威胁。本文以海州湾2004～2006年每年5～10月赤潮实测资料和同期的水文气象资料为研究对象,首先通过因子分析的方法降低模型的维数,在17种与赤潮生消过程有密切联系的环境要素的基础上提取出营养盐、温度、耗氧量和pH值、水动力、降水等共计7个特征因子。然后对因子分析的结果结合赤潮状况利用支持向量机建立赤潮状况判别模型,采用径向基函数(RBF)作为核函数,并通过选择合适的模型参数组合,取得了90%以上的平均判别正确率,可以为沿海生产和管理部门发布赤潮预警信息提供依据。

Red tides was increasingly observed in Haizhou gulf and considered as a serious environmental problem from the beginning of the new century. In present paper the authors analyzed the data of red tides, hydrology and meteorology collected during May to October of each year from 2004 to 2006 ; distill 7 characteristic factors from 17 original environmental elements to reduce the dimension of the model by means of factor analysis. The identification model of red tides was established by the method of support vector machine with these characteristic factors. The maximum accuracy arrived at 92.06% by choosing radial--basis function as the kernel function and appropriate combination of parameters. The results show that the model is efficient enough to predict the occurrence of red tide.