地下水埋深预测对于灌区农业生产、水土资源合理利用和生态环境保护等具有重要指导价值与作用。地下水埋深是一个受多种因素影响的多层次复杂系统，其演变具有不确定性、随机性、模糊性和非平稳性。基于EEMD较强的处理非线性问题能力和Elman 网络具有适应时变和动态记忆的优点，构建了基于EEMD与Elman神经网络的地下水预测耦合模型，并将其应用于人民胜利渠灌区地下水埋深预测中。研究结果表明：基于EEMD和Elman神经网络耦合模型预测结果的最大相对误差为2.91%，最小相对误差为0.04%，预测合格率为100%，该耦合模型对人民胜利渠灌区地下水埋深的预测精度要高于单一的Elman模型和BP模型。另外，模型在某种程度上可揭示灌区地下水时间序列的演变机制与因素，且计算简单、思路清晰，为地下水埋深预测提供了一种新的途径。
Abstract
Groundwater depth prediction plays an important role in agricultural production, rational utilization of land and water resources, and ecological environment protection. Groundwater depth is a multi-level complex system, which is influenced by many factors, further more, the groundwater’s evolution has the characteristics of uncertain, random, fuzzy and unstable. Based on the strong ability of EEMD to deal with nonlinear problem and the Elman network has the advantages of adaptive time and dynamic memory, a groundwater prediction coupling model based on EEMD and the Elman neural network is proposed, and applied to groundwater depth prediction in people’s victory canal irrigation district. The results show that the maximum relative error, the minimum relative error and the qualified rate are 2.91%, 0.04% and 100%, respectively, and the prediction accuracy of the coupling model is much higher than single BP model and Elman model. In addition, to some extent, the model can reveal the evolution mechanism and factors of groundwater time series, further more, the model is simple in calculation and clear in thinking, which provides a new way for groundwater depth prediction.
关键词
EEMD /
Elman 网络 /
地下水埋深 /
预测 /
人民胜利渠灌区
{{custom_keyword}} /
Key words
EEMD /
Elman network /
Groundwater depth /
Prediction /
people＇s victory canal irrigation district
{{custom_keyword}} /
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) 。
{{custom_fund}}
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] 乔雨, 梁秀娟, 王宇博.两种模型在地下水埋深预测中的应用及对比研究[J].节水灌溉, 2014, 0(3):45-53
[2] Zhang, N；Xiao, C； Liu, B.Groundwater depth predictions by GSM, RBF, and ANFIS models: a comparative assessment[J].Arabian Journal of Geosciences, 2017, 10(8):189-189
[3] Adhikary, P. & Dash, C.J...Comparison of deterministic and stochastic methods to predict spatial variation of groundwater depth[J]. Applied water science, 2017, 0(7):339-348
[4]Al-Mahallawi, K.Mania,J[J].Hani, A.. Using of neural networks for the prediction of nitrate groundwater contamination in rural and agricultural areas. Environmental Earth Sciences, 2012, 65(3):917-928
[5] Maiti, S& Tiwari, R.K.A comparative study of artificial neural networks, Bayesian neural networks and adaptive neuro-fuzzy inference system in groundwater level prediction[J].Environmental Earth Sciences, 2014, 0(71):3147-3160
[6]杨忠平, 卢文喜, 李平.时间序列模型在吉林西部地下水动态变化预测中的应用[J].水利学报, 2005, 36(12):1475-1479
[7]沈冰, 刘敏, 黄领梅.灰色自记忆模型及其在新疆和田地下水埋深预测中的应用[J].西北农林科技大学学报自然科学版, 2006, 34(11):223-226
[8]李荣峰, 沈冰, 张金凯.考虑周期性变化的地下水埋深预测自记忆模型[J].农业工程学报, 2005, 21(7):34-37
[9] Lapedes, A; Farber, R.Nonlinear Signal Processing Using Neural Network: Prediction and System Modeling[J].IEEE international conference on neural networks, 1987, 无(无):无-无
[10]Wu, Z.H.,Huang,NE.,Long,S.R. On the trend,detrending,and the variability of nonlinear and nonstationary time series.[J].Transactions of the Chinese Society of Agricultural Engineering, 2007, 104(38):14889-14894
[11] Huang, N,E., Shen, Z., Long, S.R.The empirical mode decomposition and the Hilbert spectrum for nonlinear and non-stationary time series analysis[J].Proc Royal Society Lond A, 1998, 0(454):903-995
[12]姚欣明, 陈元芳, 顾圣华.模型在降水预测中的应用[J].水电能源科学, 2014, 32(12):11-13
[13]Flandrin P, Rilling G, Gon?alves P.Empirical mode decomposition as a filterbank[J].IEEE Signal Proc Lett, 2004, 11(2):112-114
[14] Elman J L.Finding structure in time [J].Cognitive Science, 1990, 0(14):179-211
[15] 何玉婉.基于 Elman 神经网络的高速公路入口匝道预测控制仿真研究 [D]. 成都: 西南交通大学信息科与技术学院, 2008: 20?42.
[16] 石黄霞, 何颖, 董晓红.基于Elman神经网络的电力负荷预测模型研究[J].工业仪表与自动化装置, 2013, 0(1):85-87
PDF(1357 KB)
