基于PSO-SVM与ACO-SVM的沿桩长基桩轴力分布预测  

Forecasting of Pile Axial Force Distribution along Depth of Pile Based on PSO-SVM and ACO-SVM

作　　者：黄伟杰[1] 陈志坚[1] 

机构地区：[1]河海大学地球科学与工程学院,南京市210098

出　　处：《勘察科学技术》2015年第5期1-4,共4页Site Investigation Science and Technology

基　　金：国家十一五科技支撑资助项目(2006BAG04B05);973资助项目(2002CB412707)

摘　　要：由于苏通大桥群桩基础的超长桩在不同断面的轴力沿桩长分布具有复杂非线性的特点,为了实现对不同高程断面的轴力进行预测,引入粒子群算法与蚁群算法对支持向量机进行优化,在考虑断面高程、潮位、水温与风速等影响因素的基础上,建立沿桩长基桩轴力分布的优化SVM预测模型。研究表明,PSO-SVM与ACO-SVM模型比传统SVM在预测精度、模型稳定性与泛化能力方面有更好的表现,在超长桩不同高程断面的轴力预测中,具有一定的工程应用价值。Due to the complicated nonlinear characteristics of axial force distribution along pile length in different section of super-long piles in pile foundation of Sutong bridge,in order to forecast the axial force in different section in different height section,the particle swarm algorithm and ant colony algorithm are introduced to optimize the support vector machine. Based on considering the factors of elevation of section,tide level,water temperature and speed of wind and so on,the optimized SVM prediction model of pile axial force distribution along the depth of pile is built. The research shows that,compared with the traditional SVM model the PSO-SVM model and ACO-SVM model have better performance in prediction accuracy,model stability and generalization ability. It has certain engineering application value in the axial force prediction of different elevation section of super-long pile.

关 键 词：超长桩 粒子群算法 蚁群算法 支持向量机 轴力预测 

分 类 号：U443.15[建筑科学—桥梁与隧道工程]