最小二乘支持向量机模型在径流预报中的应用

The Least Square Support Vector Machine(LS-SVM) Model Applies in the Prediction of the Runoff

为解决径流中长期预报模型精度较低的问题,提出了一种基于混沌时间序列的最小二乘支持向量机预测方法,该方法依据相空间重构技术以及最小二乘支持向量机模型(以下简称LS_SVM)。利用该模型对黄河三门峡1919-1989年的71年实测年径流量进行预测。经过反复计算分析表明LS_SVM模型在径流中的应用具有较快的计算速度、较好的泛化能力以及较高的预测精度。

To solve the problem of low runoff and long-term forecasting accuracy of the model proposed in this paper, a method of least squares support vector machine is put forward on the basis of chaotic time series prediction. This method is based on phase space reconstruction and least squares support vector machine model （the referred LS_SVM）. The model is used to predict the Sanmenxia 71 years from 1919 to 1989 measured annual runoff. Analyses show that after repeated calculation LS_SVM model in the runoff com- putation speed, good generalization and high prediction accuracy.