考虑渗流与应变耦合的土坝渗透参数识别方法  

Identification of seepage parameters of earth dams based on coupling analysis of strain and seepage flow

作　　者：苗丽[1] 陈家模 

机构地区：[1]郑州大学水利与环境学院,河南郑州450002 [2]机械工业第六设计研究院,河南郑州450007

出　　处：《人民长江》2011年第3期87-90,共4页Yangtze River

摘　　要：土石坝在长期运行过程中,坝体结构形态会不断进行调整,因此对大坝渗透特性进行反演时考虑流固耦合效应是必要的。通过对大坝应力场和渗流场的耦合机理分析,研究了渗流场与应变场的耦合效应。将遗传算法和神经网络相结合,用遗传算法优化神经网络连接权值,所建立的遗传神经网络具有较快的训练速度和较强的泛化能力。将流固耦合理论、BP神经网络和遗传算法三者结合起来,应用于大坝渗透特性的有关参数反演分析中。数值算例表明,基于耦合的遗传神经网络在求解大坝渗透系数反演问题中具有较高的计算效率和识别精度。The structure behaviors of an earth dam will change in a long-time operation,so it is necessary to consider the effects of fluid-solid coupling when conducting the inversion of seepage characteristics.Through analyzing the coupling mechanism of seepage field and strain field,we study the coupling effect of seepage field and strain field.Combining the genetic algorithm with artificial neural network,and through optimizing weight coefficient of artificial neural network by genetic algorithm,the newly established genetic neural network is of faster training speed and superior generating ability.A new approach that combines genetic algorithm,fluid-solid coupling and BP neural network is applied into the inversion analysis of earth dam parameters.The presented numerical example shows that the genetic neural network based coupling is of both higher computing efficiency and higher identification accuracy.

关 键 词：BP神经网络 遗传算法 反演分析 应力场 渗流场 土石坝 

分 类 号：TV641[水利工程—水利水电工程] TV314