基于混沌分析的月径流序列耦合预测模型及其应用  

Monthly runoff coupling prediction model based on chaotic analysis and its application

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：《水力发电学报》2010年第6期41-46,共6页Journal of Hydroelectric Engineering

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：针对北方地区月径流年内年际变化幅度大、单一方法难以预测以及径流时间序列预测时,非线性模型阶数难以确定的特点,提出基于混沌分析的月径流序列耦合预测模型。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,并以相空间重构为基础进行混沌分析,然后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,为径流预测提供了一条新途径。The inter-annual variation of monthly runoff in the north China regions is large,and the prediction by a single method is often difficult.It is also difficult to determine the dimension of a nonlinear model of runoff time series.Aiming at these features,a monthly runoff coupling prediction model based on chaotic analysis is developed.This model,focusing on time-frequency analysis,divides the runoff series into different frequency components,conducts a chaos analysis by using a phase-space reconstruction,and makes a SVM prediction of the frequency components.Application of the model to the monthly runoff of a reservoir upstream of Shahe tributary,the Huaihe River valley,shows satisfactory forecast results.The model provides a new way of runoff forecasting.

关 键 词：水利管理 月径流预报 耦合预测模型 相空间重构 小波分析 支持向量机 

分 类 号：TV121.4[水利工程—水文学及水资源]