Ultimate Shear Resistance of Silty Sands Improved by Stone Columns Estimation Using Neural Network and Imperialist Competitive Algorithm

Abstract

One of the most popular soil improvement methods is stone columns. So as to investigate the performance of this approach under different types of loadings, it is essential to obtain the displacement–load curve of the stone columns. In this paper, the effect of geotextile-encased stone columns has been evaluated based on 39 large scale tests in sand with different silt content. Furthermore, artificial neural network is carried on to estimate the shear resistance of the modeled stone columns by considering some factors such as fine content of bed soil, area replacement of stone columns, and normal stress on the sample. In the present research, artificial neural networks optimized by colonial competitive algorithm (ANN-ICA) were used and their results were compared with other methods. The obtained results showed that ICA-based artificial neural networks predicted lateral bearing capacity of short piles with a correlation coefficient of 0.9738 for training data and 0.9913 for test data. Moreover, the results of the model showed the superiority of ICA-based artificial neural networks compared to back-propagation artificial neural network methods.