摘要
基于生产实践对高精度中长期径流预报提出的要求,对我国海河流域两大支流之一的滹沱河上小觉水文站(岗南水库入库径流控制站)径流量进行预测研究。采用主成分分析法提取出汛期气象因子的主成分,然后采用A Trous小波提取出序列的随机项、趋势项和周期项,分别对各项建立神经网络预测模型,然后采用A Trous小波重构法得到汛期径流预测值,得到较好的预测结果,模型能很好的反映随机项、趋势项的变化,对周期项变化的响应较随机项和趋势项的差,但也具有较高的精度,能很好的反映径流的随机性和趋势性。
Based on the requirements of high-precision medium-and long-term runoff forecasting in production practice, the runoff of Xiaojiao Hydrological Station(Gangnan Reservoir Inflow Runoff Control Station) at one of the two major tributaries of Haihe River Basin in China was predicted. The principal component analysis was used to extract the principal components of the meteorological factors in the flood season. Then, the random, dynamic and periodic terms of the sequence were extracted byA Trous wavelet. The neural network prediction model was established for each item, and then A Trous wavelet reconstruction method was used. The predicted runoff value is obtained, and the better prediction results are obtained. The model can reflect the changes of random items and trend items well. The response to periodic item changes is better than the difference between random items and trend items, but it also has higher precision. It can reflect the randomness and trend of runoff very well.
作者
周娅
郭萍
杨柳
宋培培
Zhou Ya;Guo Ping;Yang Liu;Song Peipei(Guizhou Water Resources and Hydropower Survey and Design Institute,Guiyang 550002,Guizhou;CoWWege of Water Resources and Civil Engineering,China Agricultural University,Beijing 100083,China)
出处
《陕西水利》
2019年第10期37-39,42,共4页
Shaanxi Water Resources
关键词
BP人工神经网络
小觉站径流
主成分分析法
A
Trous小波分析
BP artificial neural network
Xiaojiao station runoff
principal component analysis and A Trous wavelet analysis
