期刊文献+

基于长短期记忆神经网络模型的地下水水位预测研究 预览

Groundwater Level Prediction Based on Long-short-term Memory Neural Network Model
在线阅读 免费下载
收藏 分享 导出
摘要 利用长短期记忆神经网络(LSTM)构建地下水水位预测模型,解决了传统神经网络预测模型处理时序数据时未考虑时间序列的问题,同时采用多影响变量输入的方式弥补了简单时序模型处理数据时过于依赖时间的缺点。以泰安市岱岳区满庄镇姜家园村046J地下水位监测井为例,采用2001-2016年的监测资料与相关气候数据,利用长短期记忆神经网络构建了地下水水位预测模型,以控制变量的方法确定最优参数,对该井的地下水水位进行了预测,并与单变量LSTM神经网络、BP神经网络预测模型作对比。研究结果表明:基于多变量输入的LSTM神经网络模型能够通过少量历史数据准确的预测未来地下水水位变化情况,特别是在一些资料匮乏的地区,预测误差要显著低于参与对比的预测模型,预测均方根误差仅为2.052。因此,基于多变量的LSTM神经网络模型能够作为简单有效的地下水水位预测工具,为区域水资源管理提供一定的参考。 The long-short-term memory neural network(LSTM)is used to construct the groundwater level prediction model,which solves the problem that the traditional neural network prediction model does not consider time series when processing time series data.At the same time,the model adopts the method of multi-influence variable input to solve the problem that the simple time sequence model relies too much on time in data processing.Taking the 046J groundwater level monitoring well in Jiangjiayuan Village,Manzhuang Town,Daiyue District,Tai'an City as an example,the groundwater level prediction model is constructed by using the long-term and short-term memory neural network based on the monitoring data and related climatic data from 2001 to 2016.The optimal parameters are determined by controlling variables and the groundwater level in the study area is predicted and compared with the prediction results of univariate LSTM neural network and BP neural network.The results show that:LSTM neural network model based on multi-variable input can accurately predict future groundwater level changes by a small amount of historical data;especially in some areas where data are scarce,the prediction error is significantly lower than that of the model participating in the comparison,and the prediction RMS error is only 2.052.Therefore,LSTM neural network model based on multi-variables can be used as a simple and effective tool for groundwater level prediction and provide some reference for regional water resources management.
作者 汪云 杨海博 徐建 郑梦琪 韩智昕 赵耘 赵耀 WANG Yun;YANG Hai-bo;XU Jian;ZHENG Meng-qi;HAN Zhi-xin;ZHAO Yao(Geological Exploration Institute of Shandong Zhengyuan,TaiAn 271000,Shandong)
出处 《节水灌溉》 北大核心 2019年第10期73-77,共5页 Water Saving Irrigation
基金 山东省自然资源厅,山东省地下水水源地调查评价(泰莱盆地)(鲁地环201604)。
关键词 地下水位预测 气候条件 长短期记忆神经网络 泰安市 groundwater level prediction climatic conditions long and short term memory neural network Tai'an City
作者简介 汪云(1974),女,高级工程师,主要从事水文地质、地质工程与矿产勘察研究工作。
  • 相关文献
投稿分析

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部 意见反馈