丹江口水库径流预测方法与趋势分析  

Runoff Forecasting and Trend Analysis for Danjiangkou Reservoir

作　　者：杨正祥[1] 康玲[2] 

机构地区：[1]华中科技大学数字化工程与仿真中心,湖北武汉430074 [2]华中科技大学水电与数字化工程学院,湖北武汉430074

出　　处：《水电能源科学》2011年第2期1-3,共3页Water Resources and Power

基　　金："十一五"国家科技支撑计划基金资助项目(2006BAB04A09)

摘　　要：针对径流量时间序列非线性、非平稳性的特点,基于丹江口水库1933~2001年的入库水量资料,采用紧致型小波神经网络预测水文序列,将小波基与输入向量的内积进行加权计算和反复训练,发挥小波变换良好的时频局域化性质及神经网络的自学习功能,再通过1961~2001年降水量和入库水量的对比,分析了降水和径流的变化过程。结果表明,径流量有减少的趋势。Runoff time series have characteristics of nonlinearity and non-stationary.According to the data of Danjiangkou Reservoir inflow from 1933 to 2001 years,hydrologic series are predicted by using compact wavelet neural network.The time-frequency localization of wavelet and the self-learning function of neural network are exploited in the model of wavelet neural network through weighted calculation and repeated training on the inner product of wavelet basis and input vectors.The change process of precipitation and runoff is analyzed by the data from 1961 to 2001 years.The results show that the runoff of Danjiangkou Reservoir appears decreasing trend.

关 键 词：丹江口水库 小波神经网络 中长期预报 变化趋势 MANN-KENDALL检验 

分 类 号：TV121[水利工程—水文学及水资源] P338.2[天文地球—水文科学]