地学前缘 ›› 2016, Vol. 23 ›› Issue (3): 151-155.DOI: 10.13745/j.esf.2016.03.019

• 非主题来稿选登 • 上一篇    下一篇

基于改进的数量化理论和RBF神经网络组合方法的地下水水质预测

杨平,王新民,路来君   

  1. 1. 吉林大学 地球科学学院, 吉林 长春 130026
    2. 长春工业大学 应用数学研究所, 吉林 长春 130021
  • 收稿日期:2015-04-01 修回日期:2015-07-21 出版日期:2016-05-15 发布日期:2016-05-15
  • 作者简介:杨平（1978—），男，博士研究生，数字地质科学专业。E-mail：936342317@qq.com
  • 基金资助:

    国家自然科学基金项目(51278065)

Predicting the trends of pollutant concentrations in groundwater based on the combined method of the improved quantification theory and RBF artificial neural network

YANG Ping,WANG Xinmin,LU Laijun   

  1. 1. College of Earth Science, Jilin University, Changchun 130026, China
    2. Institute of Applied Mathematics, Changchun University of Technology, Changchun 130012, China
  • Received:2015-04-01 Revised:2015-07-21 Online:2016-05-15 Published:2016-05-15
  • Contact: 王新民（1957—），男，教授，主要从事水环境系统正反问题数学模型与数值方法的研究。

PDF (PC)

5787

摘要：

文中首先运用了一种改进的数量化理论I模型作为预处理工具,对影响地下水水质的20个因子进行定性数据转换、数据降维,随后将8个重要特征因子作为RBF（径向基函数）神经网络模型的输入,进一步对监测井的采样数据进行学习、训练，揭示地下水污染质迁移转化规律。尝试用经过改进的数量化理论与RBF神经网络方法二者结合，对沈阳李官水源地研究区监测井地下水水质变化进行模拟与预测,其仿真结果覆盖了现有的绝大部分实测数据,适用范围广泛，具有一定的推广价值。

关键词: 改进的数量化理论Ⅰ, 数据降维, 特征因子, RBF神经网络

Abstract:

In this paper, an improved quantification theory I proposed by Chikio Hayashi was used as a preprocessing tool to covert quantitative data to qualitative data and to reduce data dimensionality for 20 factors impacting groundwater quality. Then 8 important characteristic factors were used as nodes of input layer in RBF Neural Networks, and RBF ANN model was created through training and learning the sampling data of monitoring well, finally migration and transformation law of pollutants were revealed. By using the combination model of the quantification theory and RBF ANN, the simulation results covered most of the existing experimental data and could forecast the dynamic changes of groundwater quality. The result is relatively accurate for a wide range and has some promotional value.

Key words: improved quantification theory I, reduce data dimensionality, characteristic factor, RBF ANN

中图分类号: