| 基于RBF神经网络的地下水动态预测 |
| |
| 引用本文： | 张宇,卢文喜,伊燕平,陈社明.基于RBF神经网络的地下水动态预测[J].节水灌溉,2012(2):64-67,70. |
| |
| 作者姓名： | 张宇 卢文喜 伊燕平 陈社明 |
| |
| 作者单位： | 1. 吉林大学环境与资源学院,吉林长春,130026
2. 吉林省宏利水土保持咨询有限公司,吉林长春,130033 |
| |
| 摘 要： | 以内蒙古自治区巴彦淖尔市金泉工业园区为例,基于园区B248号长观井2001-2008年的地下水埋深资料,首先建立了地下水埋深RBF神经网络预测模型,而后对该模型的模拟结果作误差分析,并将相应值与BP网络模型进行对比。RBF神经网络模型和BP网络模型的最大相对误差分别为9.88%和19.67%,最大绝对误差分别为0.81和1.56,均方误差分别为0.19和0.98。显然,RBF神经网络具有较高的预测精度和较强的非线性映射能力。用上述训练好的RBF神经网络模型对研究区2009-2013年平水年条件下的地下水埋深进行预测,结果表明,研究区已出现地下水位持续下降的趋势。最后,根据地下水资源保护规划方案,在逐时段压缩地下水开采量10%的情况下,研究区2025年即可恢复到2001年的地下水水位值。
|
| 关 键 词： | RBF神经网络 BP神经网络 地下水动态 预测 |
Groundwater Dynamic Prediction Based on RBFNN |
| |
| ZHANG Yu,LU Wen-xi,YI Yan-ping,CHEN She-ming.Groundwater Dynamic Prediction Based on RBFNN[J].Water Saving Irrigation,2012(2):64-67,70. |
| |
| Authors: | ZHANG Yu LU Wen-xi YI Yan-ping CHEN She-ming |
| |
| Institution: | 1(1.College of Environment and Resources,Jilin University,Changchun 130026,Jilin Province,China; 2.Jilin Hongli Soil and Water Conservation Consulting Co.,Ltd,Changchun 130033,Jilin Province,China) |
| |
| Abstract: | Taking Jinquan Industrial Park of Bayannaoer City,Inner Mongolia as an example,RBFNN model,which could be used to predict depth to the water table was built on the data from 2001 to 2008 of long-term monitoring well B248 in the park at first.Error analysis of the simulation results was performed and comparisons to BPNN model were made.The maximum relative errors of RBFNN model and BPNN model are 9.88% and 19.67%,the maximum absolute errors of them are 0.81 and 1.56,MSE of them are 0.19 and 0.98.Obviously,RBFNN has higher accuracy,efficiency and greater nonlinear reflection ability.Using the trained RBFNN mentioned above to predict the depth to the water table in conditions of normal year from 2009 to 2013,the results show that regional groundwater level had a trend of continual dropping.Finally,according to the plan of groundwater resources protection,in conditions of the exploitation reduced by 10% period by period,the groundwater level of the study area will return to 2001 level in 2025. |
| |
| Keywords: | RBFNN BPNN groundwater dynamic prediction |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
|
