基于EEMD-BP-ANN模型的澄碧河年径流量预测研究  

Research on annual runoff prediction of Chengbihe River based on EEMD-BP-ANN model

作　　者：杜懿[1] 麻荣永[1] 赵立亚 DU Yi;MA Rong-yong;ZHAO Li-ya(College of Civil Engineering and Architecture, Guangxi University, Nanning 530004, Chin)

机构地区：[1]广西大学土木建筑工程学院,南宁530004

出　　处：《广西水利水电》2018年第2期11-14,共4页Guangxi Water Resources & Hydropower Engineering

基　　金：国家自然科学基金项目(51369005);广西防灾减灾与工程安全重点实验室系统性研究项目(2013ZDX04)

摘　　要：为了进一步提高非线性模型的预测能力,在L-M法优化的BP神经网络的基础上,结合目前应用较广的总体经验模态分解法,构建一种新的EEMD-BP-ANN非线性预测模型,并以广西澄碧河平塘水文站1963~2011年共49年的年径流量资料作为研究对象进行分析。结果表明,该新建模型预测精度较高,平均模拟相对误差仅为3.1%。该结果同时也表明,经总体经验模态分解法改进后,神经网络可以更大程度上保留原始数据的信息,从而提高模型预测精度。Aiming to improve the prediction capability of non-linear model, on the basis of BO neural network opti-mized by L-M method, a new EEMD-BP-ANN nonlinear forecast model is set up combining with EEMD method.The annual runoff data from 1963 to 2011 of Pingtang Hydrological Station on Chengbihe River in Guangxi werestudied with this model. The results of study demonstrate that this new model has fairly high prediction accuracywith the average relative error of simulation being 3.1% only;and the neural network improved by EEMD method isable to retain the information of original data to a greater extent and improve the prediction accuracy.

关 键 词：澄碧河 年径流量 神经网络 总体经验模态分解 预测精度 

分 类 号：TV121.7[水利工程—水文学及水资源]