基于小波分析的支持向量机径流预测模型及应用  

Runoff Prediction Model and Its Application Based on Wavelet Analysis and Support Vector Machine

作　　者：马细霞[1] 穆浩泽[1] 

机构地区：[1]郑州大学水利与环境学院,郑州450001

出　　处：《灌溉排水学报》2008年第3期79-81,共3页Journal of Irrigation and Drainage

基　　金：河南省自然科学基金资助项目(0411050800);河南省杰出青年科学基金资助项目(512002500)

摘　　要：针对径流年内、年际变化幅度大、单一方法难以预测的特点,提出基于小波分析的支持向量机径流预测模型。该模型从时频分析角度出发,把月径流序列分解成不同的频率成分,分别采用支持向量机进行预测。以淮河支流沙河上游某水库月径流预测为研究实例,得出了较满意的预测结果。通过与其它方法预测结果的对比分析,验证了模型的有效性,为径流预测提供了一条新途径。In allusion to these features that runoff interannual variations is very large and runoff is predicted difficultly using single method,this article presents runoff prediction model based on wavelet analysis and Support Vector Machine（SVM）. From the perspective of time-frequency analysis, runoff series can be decomposed into different frequency components, so runoff prediction model uses SVM to predict the frequency components respectively. This model is used to forecast monthly runoff of a reservoir which is in the upstream of Shahe tributary, Huaihe River, we obtained satisfactory forecast results. By comparing this method with the other forecast models, the feasibility of this model for reservoir monthly runoff forecast is proved, and this model provides a new way for runoff forecast.

关 键 词：小波分析 支持向量机 径流 

分 类 号：P333.1[天文地球—水文科学]