Numerical modeling and neural networks to identify model parameters from piezocone tests: II. Multi‐parameter identification from piezocone data

This paper completes the study presented in the accompanying paper, and demonstrates a numerical algorithm for parameter prediction from the piezocone test (CPTU) data. This part deals with a development of neural network (NN) models which are able to map multi‐variable input data onto typical geotechnical characteristics and constitutive parameters of the modified Cam clay model, which has been applied in this study. The identification procedure is designed for the coupled hydro‐mechanical boundary value problem in normally‐and lightly overconsolidated clayey soils including partially drained conditions that may appear during cone penetration. The NN models are trained with pseudo‐experimental measurements derived with the aid of the numerical model of the piezocone test, presented in the accompanying paper. Different input configurations containing CPTU measurements and some complementary data are studied with respect to the accuracy of predicted parameter values. Finally, the performance of the developed NN predictors is tested with field CPTU data which are derived from three well‐documented characterization sites, and the obtained predictions are compared with benchmark laboratory results. Copyright © 2011 John Wiley & Sons, Ltd.     

Refinement of the Hardening Soil model within the small strain range

Acta Geotechnica - The popularity of the elasto-plastic Hardening Soil (HS) model is based on simple parameter identification from standard testing and empirical formulas. The HS model is...     

A combined neural network/gradient‐based approach for the identification of constitutive model parameters using self‐boring pressuremeter tests

This paper presents a numerical procedure of material parameter identification for the coupled hydro‐mechanical boundary value problem (BVP) of the self‐boring pressuremeter test (SBPT) in clay. First, the neural network (NN) technique is applied to obtain an initial estimate of model parameters, taking into account the possible drainage conditions during the expansion test. This technique is used to avoid potential pitfalls related to the conventional gradient‐based optimization techniques, considered here as a corrector that improves predicted parameters. Parameter identification based on measurements obtained through the pressuremeter expansion test and two types of holding tests is illustrated on the Modified Cam clay model. NNs are trained using a set of test samples, which are generated by means of finite element simulations of SBPT. The measurements obtained through expansion and consolidation tests are normalized so that NN predictors operate independently of the testing depth. Examples of parameter determination are demonstrated on both numerical and field data. The efficiency of the combined parameter identification in terms of accuracy, effectiveness and computational effort is also discussed. Copyright © 2008 John Wiley & Sons, Ltd.     

Optimization framework for calibration of constitutive models enhanced by neural networks

A two‐level procedure designed for the estimation of constitutive model parameters is presented in this paper. The neural network (NN) approach at the first level is applied to achieve the first approximation of parameters. This technique is used to avoid potential pitfalls related to the conventional gradient‐based optimization techniques, considered here as a corrector that improves predicted parameters. The feed‐forward NN (FFNN) and the modified Gauss–Newton algorithms are briefly presented. The proposed framework is verified for the elasto‐plastic modified Cam Clay model that can be calibrated based on standard triaxial laboratory tests, i.e. the isotropic consolidation test and the drained compression test. Two different formulations of the input data to the NN, enhanced by a dimensional reduction of experimental data using principal component analysis, are presented. The determination of model characteristics is demonstrated, first on numerical pseudo‐experiments and then on the experimental data. The efficiency of the proposed approach by means of accuracy and computational effort is also discussed. Copyright © 2008 John Wiley & Sons, Ltd.