摘要
地下水埋深受多种因素的影响,其演变具有趋势性、突变性与非平稳性。小波分解通过将时间序列分解成高频与低频分量来减少序列的非平稳性,其在处理非平稳信号方面具有明显的优势;Elman网络对非线性问题具有适应时变和动态记忆的优点。基于小波分解与Elman网络的优点,提出了一个新的地下水埋深预测耦合模型,并将其应用到人民胜利渠地下水埋深预测中;为验证模型的可靠性,将其预测结果分别与CEEMD-Elman模型和BP网络模型的预测结果进行了对比。结果表明:基于小波分解-Elman网络预测模型的最大相对误差为30.7%,最小相对误差为0.2%,平均相对误差为2.2%,预测效果较好;随着小波分解-Elman网络预测模型分解层数的增加,子分量与地下水埋深的相关性有增强的趋势,序列的平稳性越来越好;CEEMD-Elman模型和BP网络模型的平均相对误差分别为8.1%、3.8%,说明小波分解-Elman模型的精度最高。
The buried depth of groundwater is affected by many factors,and its evolution has tendency,mutability and non-stability.The non-stationary nature of sequences is reduced by wavelet decomposition with decomposing time series into high frequency and low frequency components.It has obvious advantages in dealing with non-stationary signals.Elman network has the advantages of adapting to time-varying and dynamic memory for nonlinear problems.Based on the advantages of wavelet decomposition and Elman network,a new groundwater depth prediction coupling model was proposed and applied to the prediction results of the buried depth of groundwater in the People′s Shengli Canal.In order to verify the reliability of the model,the prediction results were compared with those of the CEEMD-Elman model and the BP network model.The results show that the maximum relative error of the prediction model based on wavelet decomposition and Elman network is 30.7%,the minimum relative error is 0.2%,the average relative error is 2.2%,and the prediction effect is sound.With the increase of wavelet decomposition and the number of decomposition layers of the Elman network prediction model,the correlation between sub-components and groundwater depth has increased,and the stability of the sequence is getting better and better.The average relative errors of the CEEMD-Elman model and the BP network model are 8.1%and 3.8%,respectively,indicating that the wavelet decomposition-Elman model has the highest accuracy.
作者
张先起
牛昂
宋超
ZHANG Xianqi;NIU Ang;SONG Chao(x09(1.School of Water Conservancy,North China University of Water Resources and Electric Power,Zhengzhou 450046,China;Collaborative Innovation Center of Water Resources Efficient Utilization and Protection Engineering,Zhengzhou 450046,China)
出处
《华北水利水电大学学报（自然科学版）》
2020年第1期1-7,共7页
Journal of North China University of Water Resources and Electric Power：Natural Science Edition
基金
国家自然科学基金项目(U1304511)
河南省国际科技合作项目(152102410052)
关键词
小波分解
ELMAN网络
预测
灌区
地下水埋深
wavelet decomposition
Elman network
prediction
irrigation area
the buried depth of groundwater
