Introducing a New Artificial Neural Network Model for prediction of the Pressuremeter Modulus in soils of Tehran
محورهای موضوعی : Engineering geology
Shahin Razavi
1
(Department of Mining Engineering, Science and Research Branch, Islamic Azad University, Tehran, Iran)
Kamran Goshtasbi
2
(Department of Mining Engineering, Tarbiat Modares University, Tehran, Iran)
Ali Noorzad
3
(Faculty of Civil, Water and Environmental Engineering, Shahid Beheshti University, Tehran, Iran)
Kaveh Ahangari
4
(Department of Mining Engineering, Science and Research Branch, Islamic Azad University, Tehran, Iran)
کلید واژه: Artificial Neural Network, In-situ test, Soil deformation modulus, Pressuremeter,
چکیده مقاله :
Pressuremeter is one of the most reliable in-situ tests in geotechnical engineering. Soil deformation modulus has been related empirically to the pressuremeter modulus (E ) obtained from the pressurevolume change curve from this test. In general, the pressuremeter test is time-consuming and costly that requires experienced operators. Various parameters might also affect the test results. With these limitations, it is necessary to introduce equations and models for indirect determination of the E. Artificial neural network (ANN) is a very useful technique for modeling complex relationships between input and output data sets. The ANN models often produce more accurate results compared with the linear regression methods. The main purpose of this research is to introduce a new ANN model for prediction of the EPM. The data used in this research is taken from 41 pressuremeter tests in soils of Tehran. In order to estimate EPM, parameters such as grain size distribution, depth of test, and moisture content are considered as input (independent) variables. The coefficient of determination (R2) for the training, validation, and test data sets were 0.736, 0.906, and 0.801, respectively. Acceptable correlations and errors of network predictions in comparison with the actual values of EPM show the accuracy and efficiency of the designed model. Sensitivity analysis revealed that the grain size distribution is the most effective parameter among the variables on the EPM.
