欢迎访问中国科学院大学学报,今天是

中国科学院大学学报 ›› 2011, Vol. 28 ›› Issue (1): 51-56.DOI: 10.7523/j.issn.2095-6134.2011.1.008

• 论文 • 上一篇    下一篇

基于BP算法的地下水模拟中 加速因子的确定

马荣1, 刘继朝1, 石建省1, 王虎2   

  1. 1. 地质科学院水文地质环境地质研究所,石家庄 050803;
    2. 中国石油大学(北京),北京 102249
  • 收稿日期:2010-03-09 修回日期:2010-05-24 发布日期:2011-01-20
  • 基金资助:

    国家重点基础研究发展规划(2010CB428800)和中国地质科学院水文地质环境地质研究所项目(SK201015)资助 

Determination of the acceleration factor in groundwater simulation process through BP algorithm

MA Rong1, LIU Ji-Chao1, SHI Jian-Sheng1, WANG Hu2   

  1. 1. Institute of Hydrogeology and Environmental Geology,CAGS, Shijiazhuang 050803,China;
    2. China University of Petroleum(Beijing), Beijing 102249,China
  • Received:2010-03-09 Revised:2010-05-24 Published:2011-01-20

摘要:

在大型线性方程组的超松弛迭代法求解中,加速因子经常难以确定.应用BP神经网络对其进行训练学习,经过对比分析,得到最佳模型,应用该模型可快速确定加速因子.将该方法应用于石家庄市栾城水文试验基地,计算结果表明,BP人工神经网络有效地解决了地下水数值模拟中加速因子难以确定的问题.

关键词: 地下水数值模拟, 加速因子, 超松弛迭代法, 神经网络

Abstract:

In applying the successive over-relaxation iteration method to solve large-scale linear equations,one often has difficulties in determining the acceleration factor.Through training and learning using BP neural network and comparative analyses, we obtained a good model,which could be used for fast determination of the acceleration factor.We used it in the Luancheng hydrology experimental base in Shijiazhuang,and the results show that BP artificial neural network has been successfully used in solving the difficult problem, determination of the acceleration factor in the groundwater numerical simulation process.

Key words: groundwater numerical simulation, acceleration factor, successive over-relaxation iteration method, neural network

中图分类号: