Intelligent Prediction for Side Friction of Large-Diameter and Super-Long Steel Pipe Pile Based on Support Vector Machine

Abstract

In recent years, more and more large-diameter and super-long steel pipe piles are applied in engineering project. But people just know little about the bearing characteristics of super-long piles as it is very difficult to study such type of super-long piles in the laboratory and the accumulated test data of super-long piles in actual projects is very few restricted by test conditions and test cost. In engineering work, design value of bearing capacity of large-diameter and super-long piles is still referred to the calculation theory of ordinary pile that cannot take into account engineering security and economic simultaneously. In this paper, SVM-Q which is an intelligent algorithm based on Support Vector Machines is developed for predicting side friction of large-diameter and super-long steel pipe pile. Result shows that the side friction of longer large-diameter and super-long steel pipe piles with similar bearing characteristics can be effectively predicted by the SVM-Q algorithm after fully learning enough side friction data samples of the limited testing piles with gradually larger length, and boundary length of super-long steel pipe pile in this actual engineering could be qualitatively judged by comparing predictive data with the measured data. This method is very meaningful for initiative predicting the bearing capacity of large-diameter and super-long steel pipe piles in the case that there is no suitable calculation method. The predictive bearing capacity also can be adopted to verify the bearing capacity of large-diameter and super-long steel pipe piles that donot be field-tested by static load tests in actual projects.