支持向量机及其在径流预测中的应用

Support Vector Machine Method and Its Application to Prediction of Runoff

给出了支持向量机方法(SVM)的思路、特点及关键之处,探讨了SVM在径流预测中的可能性,并与基于遗传算法的门限回归模型(TR)进行了对比分析。径流预测实例分析表明,在拟合阶段,SVM模型要好于TR模型;在预留检验阶段,SVM模型与TR模型接近。同时SVM模型适合于小样本情况且能达到全局最优。SVM模型用于径流预测是可行的、优越的。

The main idea, features and key points of support vector machine （SVM） have been introduced in this paper. Then prediction of runoff has been studied with SVM technique, and compared with traditional threshold regression model （TR） based on genetic algorithm. The results of case studies have shown that the SVM method is functional and feasible.