Principal Component Analysis and Neural Networks for Predicting the Pile Capacity Using SPT

Abstract

A neural network is, in essence, an attempt to simulate the brain. Neural network theory revolves around the idea that certain key properties of biological neurons can be extracted and applied to simulations, thus creating a simulated (and very much simplified) brain. The first important thing to understand then is that the components of an artificial neural network are an attempt to recreate the computing potential of the brain. This famous network memorizes information by a process of training, to this effect the theory of artificial neural network is developed and is applied in several fields of sciences. The geotechnical domain is among them and in particular the resolution of problems of which parameters that govern them have an uncertain character, as the case of the prediction of the pile capacity. For it we collected 120 cases of the literature, sweeping a variety of sites through the world. The model conceived by an iterative process that is, the retropropagation was validated by experimental tests and was compared with the values predicted by four of the most commonly used traditional methods. In this paper, the developed neural network model is based on the principal component analysis approach (PCA) for data analysis in the aim to improve the generalization process. The results indicate that the ANN model is able to accurately predict the capacity in several cases, including the experiments on model piles. The PCA technique shows the efficiency in the variable analysis in order to determine their relative contribution on the pile capacity and improve the generalization capacity. This study is limited for the driven piles.

Prediction of Compression Index of Soft Soils from the Brazilian Coast Using Artificial Neural Networks and Empirical Correlations

“…Further studies on the application of ANN in geotechnical engineering include the prediction of properties like the hydraulic conductivity in clays (Goh, 1995), the optimum water content and the corresponding maximum dry density of the soil (Najjar et al, 1996) and the residual friction angle prediction of clays (Das & Basudhar, 2008). ANNs were also used for soils settlement estimation (Nejad et al, 2009;Benali et al, 2013) and shear strength parameter prediction (Khanlari et al, 2012).…”