基于小波广义回归神经网络耦合模型的月径流预测

发布时间：2018-01-18 13:23

本文关键词：基于小波广义回归神经网络耦合模型的月径流预测　出处：《水力发电学报》2016年05期 　论文类型：期刊论文

【摘要】：针对中长期水文预报方法预测结果精度低的问题,将离散小波变换(DWT)与广义回归神经网络(GRNN)耦合,建立了月径流预测模型。通过DWT处理将原始月径流序列分解重构为确定性成分和随机性成分两个分量,对两个分量的GRNN模型预测结果叠加作为预测值的方法称为WGRNN1模型。将WGRNN1模型与剔除随机序列的GRNN模型(WGRNN2)和不进行离散小波变换的GRNN模型结果进行对比,采用平均绝对误差(MAE)、确定性系数(DC)和相关系数(R)为模型评价指标。将模型应用于黑河干流莺落峡站的月径流预测,结果表明:模型WGRNN2的评价指标优于WGRNN1,且这两个模型预测效果都优于GRNN模型。说明与离散小波变换的耦合可以提高GRNN模型对月径流的预测精度,同时剔除随机成分的小波广义回归神经网络模型有更好的预测效果,可应用于实际生产。
[Abstract]:In order to solve the problem of low accuracy of the prediction results of the medium and long term hydrological forecasting methods, the discrete wavelet transform (DWT) and the generalized regression neural network (GRNN) are coupled. The monthly runoff prediction model is established and the original monthly runoff series is decomposed into deterministic components and stochastic components by DWT processing. The method of superimposing the prediction results of two components of GRNN model as the prediction value is called WGRNN1 model. The WGRNN1 model and the GRNN model which excludes random sequences are combined with WGRN2). The results are compared with the results of GRNN model without discrete wavelet transform. The mean absolute error (mae), deterministic coefficient (DC) and correlation coefficient (R) were used as the evaluation index. The model was applied to the monthly runoff prediction of Yingluoxia station in the main stream of Heihe River. The results show that the evaluation index of model WGRNN2 is better than that of WGRNN1. These two models are better than GRNN model, which shows that the coupling with discrete wavelet transform can improve the accuracy of monthly runoff prediction by GRNN model. The wavelet generalized regression neural network model, which excludes the random components, has better prediction effect and can be used in practical production.
【作者单位】：西北农林科技大学水利与建筑工程学院;
【基金】：国家自然科学基金(91425302;51279166)
【分类号】：TV121
【正文快照】： 0引言中长期径流预测可为流域水资源的规划利用和运行管理提供依据。对径流的准确模拟和预测一直是国内外水文学者关注和研究的重点问题之一,尽管已经发展了很多预测模型和方法,但由于径流形成过程受到降水量、下垫面和人类活动等诸多因子的影响,其变化规律具有复杂且高度非线

【参考文献】