Ali Ghorbani

Estimating shear wave velocity of soil deposits using polynomial neural networks: Application to liquefaction

Geophysical and geotechnical field investigations have introduced several techniques to measure in-situ shear wave velocity of soils. However, there are some difficulties for the easy and economical use of these techniques in the routine geotechnical engineering works. For the soil deposits, researchers have developed correlations between shear wave velocity and SPT-N values. In the present study, a new database containing the measured shear wave velocity of soil deposits have been compiled from the previously published studies. Using polynomial neural network (PNN), a new correlation has been subsequently developed for the prediction of shear wave velocity. The developed relationship shows an acceptable performance compared with the available relationships. Three examples are then presented to confirm accuracy and applicability of the proposed equation in the field of liquefaction potential assessment.

Incremental communication for multilayer neural networks: error analysis

Artificial neural networks (ANNs) involve a large amount of internode communications. To reduce the communication cost as well as the time of learning process in ANNs, we earlier proposed (1995) an incremental internode communication method. In the incremental communication method, instead of communicating the full magnitude of the output value of a node, only the increment or decrement to its previous value is sent to a communication link. In this paper, the effects of the limited precision incremental communication method on the convergence behavior and performance of multilayer neural networks are investigated. The nonlinear aspects of representing the incremental values with reduced (limited) precision for the commonly used error backpropagation training algorithm are analyzed. It is shown that the nonlinear effect of small perturbations in the input(s)/output of a node does not cause instability. The analysis is supported by simulation studies of two problems. The simulation results demonstrate that the limited precision errors are bounded and do not seriously affect the convergence of multilayer neural networks.

A novel solution for ground reaction curve of tunnels in elastoplastic strain softening rock masses

AbstractGround Reaction Curve (GRC) is one of the most important elements of convergence-confinement method generally used to design tunnels. Realistic presentation of GRC is usually assessed based on the advanced rock strength criteria, also, rock mass behavior (including plasticity and softening treatments). Since taking these parameters into account is not simply possible for practitioners and needs complicated coupled theoretical-numerical solutions, this paper presents a simple novel approach based on Evolutionary Polynomial Regression to determine GRC of rock masses obeying both Mohr-Coulomb and Hoek-Brown criteria and strain softening behaviors. The proposed models accurately present support pressures based on radial displacement, rock mass strength and softening parameter (determination coefficient of 97.98% and 94.2% respectively for Mohr-Coulomb and Hoek-Brown strain softening materials). The accuracy of the proposed equations are approved through comparing the EPR developed GRCs with the ground...

Simplified dynamic analysis to evaluate liquefaction-induced lateral deformation of earth slopes: a computational fluid dynamics approach

Lateral deformation of liquefiable soil is a cause of much damage during earthquakes, reportedly more than other forms of liquefaction-induced ground failures. Researchers have presented studies in which the liquefied soil is considered as viscous fluid. In this manner, the liquefied soil behaves as non-Newtonian fluid, whose viscosity decreases as the shear strain rate increases. The current study incorporates computational fluid dynamics to propose a simplified dynamic analysis for the liquefaction-induced lateral deformation of earth slopes. The numerical procedure involves a quasi-linear elastic model for small to moderate strains and a Bingham fluid model for large strain states during liquefaction. An iterative procedure is considered to estimate the strain-compatible shear stiffness of soil. The post-liquefaction residual strength of soil is considered as the initial Bingham viscosity. Performance of the numerical procedure is examined by using the results of centrifuge model and shaking table tests together with some field observations of lateral ground deformation. The results demonstrate that the proposed procedure predicts the time history of lateral ground deformation with a reasonable degree of precision.