摘要
传统方法在计算含水层渗透系数的过程中通常面临着不均匀系数C难以确定、对研究区水文地质条件较为依赖等困难。通过分析和概括内分泌系统独特的信息处理功能,构建了人工内分泌网络模型。在详细介绍模型算法和计算步骤的基础上,将其应用到华北平原滹沱河冲洪积扇前缘,根据孔隙度及粒度分布特征对含水层渗透系数进行预测,并与传统的经验公式Beyer法和Slichter法进行对比。计算结果表明:Beyer法的预测精度最低,其相对误差主要集中于0.078～1.342;Slichter法的预测精度有所提高,但对于渗透系数极端值的预测仍存在较大误差,相对误差主要集中于0.046～0.643;人工内分泌网络模型具有较高的预测精度,其相对误差主要集中于0.006～0.420。与传统经验公式相比,人工内分泌网络模型具有较高的计算精度和良好的通用性,且能够直接计算出含水层的垂向渗透系数,无需后期的数据验证。
Although various methods can be used to determine hydraulic conductivity of aquifer,each method has faults that limit application,such as the results mainly dependent on the geometry and hydraulic boundaries of the streambed;The accurate value of coefficient C is difficult to obtain.The authors proposed a new method-Artificial Endocrine Network Model（AENM）,and described its algorithm principle.In order to verify the validity of AENM,the traditional empirical formulas and AENM were applied in alluvial-pluvial fan front in North China Plain（NCP） to calculate hydraulic conductivity according to grain-size distribution and porosity.The results indicate that the Beyer formula was less accuracywith a relative error of 0.078-1.342,as to the the Slichter method,the prediction accuracy rate was increased,but the forecasting error of extreme value was still larger with a relative error of 0.046-0.643;The prediction accuracy of AENM was higher,the relative error of 0.006-0.420.The AENM method has several advantageous over traditional methods which has not only higher calculation precision and stronger versatilitybut also could directly estimate vertical hydraulic conductivity without data transformation too long sentence to be revised.Therefore,the AENM method could be successfully used to study the hydrogeology parameter.
出处
《吉林大学学报（地球科学版）》
EI
CAS
CSCD
北大核心
2013年第3期914-921,共8页
Journal of Jilin University：Earth Science Edition
基金
国家"973"计划项目(2010CB428800)
关键词
人工内分泌网络模型
渗透系数
孔隙度
水文地质
含水层
artificial endocrine network model
hydraulic conductivity
porosity
hydrogeology
aquifer
