prediction of ultimate bearing capacity of axially loaded piles using a support vector machine

bearing capacity prediction of axially loaded piles is one of the most important problems in geotechnical engineering practices, with a wide variety range of methods which have been introduced to forecast it accurately. machine learning methods have been reported by many contemporary researches with some degree of success in modeling geotechnical phenomena. in this study, a fairly new machine learning method known as support vector machine (svm) has been used to develop a model to approximate the ultimate bearing capacity of axially loaded piles, based on cone penetration test (cpt) data. the utilized dataset obtained from published literature contains full scale static load test and cpt results and pile geometry for each sample. aditionally, sensitivity analysis of the model respect to each input parameter has been investigated. finally, a comparison between actual values and predicted bearing capacity confirms efficiency of the developed model.