摘要
提出将EMD与LS-SVM模型相耦合的新的径流中长期预测方法,采用EMD分解年径流序列,应用LS-SVM模型预测和重构IMF分量。基于岷江紫坪铺水文站年径流资料预测和检验该模型,并与单独的LS-SVM模型及BP神经网络模型比较。实例结果表明,该方法预报精度高,预测径流行之有效。
Based on least square support vector machine (LS-SVM) and empirical mode decomposition (EMD), a combination method of raid-long term runoff forecasting is proposed in this paper. The annual runoff series is decomposed by EMD method. The LS-SVM) model is used to forecast and reconstruct components of intrinsic mode function. This method is applied to annual runoff forecasting of Zipingpu Station in Minjiang River. Compared with the results obtained by LS-SVM) and BP neural network method, it shows that the proposed runoff forecasting method is effective and has high forecasting precision.
出处
《水电能源科学》
北大核心
2010年第4期11-13,共3页
Water Resources and Power
关键词
水文
经验模态分解
最小二乘支持向量机
中长期
径流预测
hydrology
empirical mode decomposition
least square support vector machine
mid and long term
runoff forecasting