Prediction of Lateral Bearing Capacity of Pile in Clay Using Support Vector Machine

A number of empirical formulas were proposed so as to reduce the time and cost involved in static approach to determine the pile capacity. Although these formulas have been widely used to predict pile capacity, it is agreed that these formulas are inaccurate due to their oversimplification of the modeling of the hammer, driving system, pile, and soil (Fragaszy et al. 1985). The support vector machine (SVM) is a relatively new artificial intelligence technique which is increasingly being applied to geotechnical problems. An important feature of the SVM is that it endeavors to discover the rules (or functions) that govern a phenomenon using only a set of data (a set of measured inputs and their corresponding outputs). Hence, there is no need to incorporate any assumptions to simplify the problem as, is often the case with many traditional methods. The previous studies indicated that these methods are more accurate compared to analytical formulas. In this paper, SVM technique is presented to predict the undrained lateral load capacity of piles in clay using the diameter of pile (D), depth of pile embedment (L), eccentricity of load (e), undrained shear strength of soil (Su) as the inputs of model. The model was developed and tested using an experimental dataset. The performance of the proposed model (SVM) was compared with ANN and those of theoretical methods of Broms, and Hansen.