摘要
分析了管涌发生的过程和影响管涌发生的因素,提出了一种预测判定管涌发生可能性的机理模型.根据机理模型从影响堤防和土石坝管涌发生的诸多复杂因素中选出既便于测量、建立观测又对管涌发生影响显著的几种因素作为系统输入,把理论机理模型和改进的BP人工神经网络模型相结合,建立预测判定堤防和土石坝中管涌发生的人工智能方法,对管涌发生的可能性因子进行了预测.并通过数据库功能,在应用中不断增加训练样本的规模,使神经网络能够学习到更全面的知识.预测结果的精度较高,表明该方法是可行的.
A hybrid model to predict and judge seepage piping occurring was presented by combining the mechanism model and the neural network model. A set of factors, corresponding with the reliable data which had significant effects on the judgment of seepage piping occurring and easy to observe and measure, were filtrated based on the mechanism model from a great number of complex and disorderly observed engineering data. This data as the effective parameters were applied into a modified BP neural network scheme to analyze the characteristics of the seepage piping occurring. The modified BP neural network model combined with a database system was designed as an artificial intelligent method to predict and judge the seepage piping occurring in embankments. The developed neural network model was applied to some practical embankments for judging and forecasting of the possibility of seepage piping failure using the collected data from a number of reservoirs and embankments. The results show that the proposed method is effective.
出处
《浙江大学学报（工学版）》
EI
CAS
CSCD
北大核心
2004年第7期902-908,共7页
Journal of Zhejiang University：Engineering Science
基金
教育部博士点基金资助项目(A50221).
关键词
堤坝
管涌
机理模型
人工神经网络
Backpropagation
Database systems
Neural networks
Seepage
Soils
