堆石体流变本构模型参数的智能反演

提出基于演化神经网络和序列二次规划算法的智能反演方法，通过实测变形资料对堆石坝流变本构模型参数进行反演分析，算例计算表明，该方法高效且具有全局收敛性。采用该方法对某堆石坝9参数流变模型参数进行反演分析，计算结果表明，利用反演的模型流变参数所计算的坝体流变变形与实测资料在发展规律及数值上均比较吻合。最后采用反演后的参数对该堆石坝后期变形进行了考虑流变效应的流变计算，预测出该堆石坝的后期变形量，预测结果合理可信。

Intelligent back analysis on parameters of creep constitutive model

An intelligent back analysis method based on evolutionary neural network(NN) algorithm and sequence quadratic programming (SQP) is proposed. The parameters of the creep constitutive model can be obtained from the back analysis according to the observed deformations of concrete face rockfill dam(CFRD) in prototype and then applied to predict the post deformation of the dam with the effect of creep taken into account. The application of the proposed method to predict the post deformation of a CFRD with height of 234m using power function creep constitutive model with nine parameters shows that the result is reasonable and credible.