摘要
利用聚类分析法将径流序列分解为若干个子径流序列 ,对这些子径流序列分别建立局部神经网络模型 ,而后把这些局部模型合并成一个混合模型。当新的信息进入该模型时 ,首先用分类器判别其类别 ,以确定用混合模型中的何种局部模型加以模拟。通过与不加分类的总体神经网络模型的模拟结果加以对比 ,结果表明这种基于径流分类的降雨 -径流模型表现出了更优良的性能 ,可以较大地提高径流模拟精度。
A runoff sequence was classified into several sub-runoff sequences with cluster analysis, and local artificial neural network (LANN) for each sub-runoff sequence was performed separately. These LANNs then was conflated into an integrated model. When a new data fed into the integrated model, a classifier will deliver the new data into different non-linear local ANN model. Comparing the performance of the new model with that of the lumped global ANN illustrated that the runoff classified local ANN rainfall-runoff model is more suitable to daily runoff forecasting.
出处
《灌溉排水》
CSCD
北大核心
2002年第4期45-48,共4页
Irrigation and Drainage
基金
河海大学创新基金项目
