岩土非线性流变本构模型的辨识算法研究

Identification method for nonlinear constitutive model of soft rock

  • 摘要: 从岩土非线性流变本构模型通式——P.Perzyna方程出发,研制了岩土工程围岩流变本构模型辨识用有限元计算程序EVP2D,并将其和均匀设计(UD)法结合,获得了用于ANN辨识模型训练用的具有丰富本构信息的全局性输入输出有效数据。尔后,设计了基于实际监测位移数据辨识围岩流变本构模型参数的ANN模型,并在matlab软件平台中研制了辨识用相关程序CYJBS3.M。有关实例验证,从待辨识参数向量UD设计、训练数据的有限元计算获取、ANN模型的训练和辨识这一整体计算过程的实现,表明了UD-FEM-ANN本构模型辨识方法的可行性。

     

    Abstract: Based on P.Perzyna equation,which is the general nonlinear rheological constitutive equation for geotechnical material,the paper develops the FEM computation program EVP2D for the identification purpose.The program computation combining the uniform design method produces a number of effective overall input output data,which contain a great deal of constitutive information and could be used to train the ANN identification model.Then an ANN model is designed based on the practical measuring displacements for the rheological constitutive model identification of the surrounding rockmass and a program CYJBS3.M for its training and identifying algorithm is also developed on the base of MATLAB software.A real application example shows that the UD-FEM-ANN identification method is successful and executive by performing its whole computation from the uniform design of the identified constitutive parameters of the surrounding rock rheological model,the FEM computation of the training data to the training and identification of the ANN model.

     

/

返回文章
返回