水利水电技术2018，Vol.49Issue(2)：43-48,6.DOI:10.13928/j.cnki.wrahe.2018.02.007
RBF神经网络在地下水动态预测中的应用
Application of RBF Neural Network for Groundwater Dynamic Prediction
摘要
Abstract
In order to predict dynamically groundwater level,the neural network model is applied to the construction of groundwater depth prediction model.Taking Changchun city as an example,the dynamic process of buried depth is simulated by radial basis function(RBF) and reverse propagation(BP) algorithm model optimization parameters within 84 sets of data from 2006 to 2012 as training sample and 36 sets of data from 2013 to 2015 as test sample,in full of approximation convergence cappcity of RBF nerve network.The two models simulate the measured buried depth dynamic process,and compare the performance.The results indicate that following parameters of the RBF neural network model and BP neural network model:the root mean square errors are 0.10 and 0.43,respectively.The maximum absolute errors are 0.44 m and 0.61 m,respectively,and the maximum relative errors are 14.60％ and 27.17％ respectively;groundwater level of Changchun city has obvious cyclic and large variation after 2015 with obvious seasonal characteristics,and the maximum depth is 5.1 m in rainless preriod but the minimum depth is 1.62 rn in wet season,obviously,the RBF model has better nonlinear mapping ability and prediction accuracy.The model can be used to predict the dynamic data of similar.关键词
地下水动态预测/RBF/BP/神经网络Key words
groundwater dynamic prediction/radial basis function(RBF)/back propagation(BP)/neural network分类
地质学引用本文复制引用
曹文洁,肖长来,梁秀娟,韩良跃,胡冰..RBF神经网络在地下水动态预测中的应用[J].水利水电技术,2018,49(2):43-48,6.基金项目
国家自然科学基金项目(41572216) （41572216）
中国地质调查局沈阳地质调查中心“长吉经济圈地质环境综合调查”项目(121201007000150012) （121201007000150012）
吉林省自然科学基金项目(20140101164JC) （20140101164JC）
