基于支持向量机的太湖入湖河流水质影响因素的研究  

Study on the key influential factors of water quality of rivers flowing into Taihu Lake based on support vector machine

作　　者：丁彦蕊[1] 孙小妹[1] 王文超[1] 孙培冬 陈蓓 

机构地区：[1]江南大学物联网工程学院,江苏无锡214122 [2]江南大学化学与材料工程学院,江苏无锡214122 [3]无锡市滨湖区环境监测站,江苏无锡214072

出　　处：《水资源与水工程学报》2011年第5期38-40,46,共4页Journal of Water Resources and Water Engineering

基　　金：国家自然科学基金项目(21001053);中央高校基本科研业务费专项资金资助(JUSRP11126);江苏省环境检测科研基金项目(0902)

摘　　要：以影响太湖入湖河流水质的23个因素值为研究对象,通过分析这些因素值随时间的变化趋势,发现随时间变化较大的因素为TN,CODMn,TP,石油类,Cd和Pb;以这些因素值为特征向量,利用支持向量机对轻度污染水质和重度污染水质进行了分类。当支持向量机的参数c为18.95、g为2.09时,局部预测率和全局预测率较高,分别为82.22%和85.71%,重度污染水质和轻度污染水质的预测率分别为70%和94.44%,这不仅表明TN,CODMn,TP,石油类,Cd和Pb是影响水体水质的主要因素,而且表明构建的支持向量机分类器适合于水质预测。This paper studied 23 factors that influenced the water quality of rivers flowing into Taihu Lake.After analyzing the factors trend with time,we found TN,CODMn,TP,petroleum,Cd and Pb changed more.Then,support vector machines were used to classify the lightly polluted water and heavily polluted water.When the parameter c is 18.95 and g is 2.09,the local prediction accuracy is 82.22% and the global prediction accuracy is 85.71%.The prediction accuracies of heavily polluted water and lightly polluted water are 70% and 94.44% respectively.This result not only showed that TN,CODMn,TP,petroleum,Cd and Pb were important factors that polluted the water but also showed the supported vector machines can be suieble to predict water quality.

关 键 词：支持向量机 影响因素 水体水质 

分 类 号：X824[环境科学与工程—环境工程]