基于支持向量机SVR的黄河凌汛预报方法  

Yellow River ice flood prediction based on SVR

作　　者：李亚伟[1] 陈守煜[1] 韩小军[1] 

机构地区：[1]大连理工大学土木水利学院,辽宁大连116024

出　　处：《大连理工大学学报》2006年第2期272-275,共4页Journal of Dalian University of Technology

基　　金：国家自然科学基金资助项目(59179376)

摘　　要：黄河内蒙段每年都有不同程度的凌汛灾害发生,准确及时的凌汛预报能够为防汛工作提供决策支持.但至今尚无一种令人满意的预测模型,为此提出一种基于支持向量机回归(SVR)的凌汛预报模型.SVR是基于统计学习理论的一种机器学习(m ach ine learn ing)方法,具有严格的理论基础,尤其是在小样本情况下,它能够利用有限的样本信息获得最好的学习效果和泛化能力.实例分析结果表明,基于SVR的凌汛预报方法具有训练速度快、泛化能力强的特点,对黄河内蒙段凌汛期封河历时预测比较准确,这对黄河凌汛防范和水资源的可持续发展具有重要意义.Ice flood occurs almost every year in Inner Mongolia reach of Yellow River, so ice flood prediction has become an important tool to support engineers to deal with ice flood problem. However, there is still not a satisfied model to deal with this problem until now. So an ice flood prediction model based on support vector machine regression （SVR） model is proposed which is a kind of machine learning model based on statistics learning theory. Especially, SVR does well in dealing with small stylebooks and can acquire good generalization ability with limited information （stylebooks）. The case study shows that SVR＇s training speed and generalization ability are better than those of ANN prediction, which is meaningful to the ice flood prevention and water resources＇ sustainable development of Yellow River.

关 键 词：支持向量机 黄河凌汛 预测 

分 类 号：P338[天文地球—水文科学]