基于GA-SVM的深水群桩基础桩顶轴力预测  

Forecasting of Axial Force in Deep-water Group Top Pile Foundation Based on GA-SVM

作　　者：陈兵[1] 黄伟杰[2] 王晓朋[1] 张文海[1] 

机构地区：[1]中国电建集团贵阳勘测设计研究院有限公司,贵阳市561110 [2]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》2016年第1期6-9,29,共5页Site Investigation Science and Technology

摘　　要：对深水群桩基桩轴力进行精确地预测是评价基础安全稳定性的重要考量依据,由于深水群桩基础的受力状态与群桩基础所处的环境呈现出复杂非线性关系。该文在某大桥运营期原型监测数据的基础上,引入遗传算法优化支持向量机模型,深入分析影响基桩轴力变化的环境因素,建立了多因素基桩轴力预测模型,并将预测结果与传统SVM模型、RBF神经网络模型进行对比。研究表明,与SVM、RBF的预测结果相比,GA-SVM模型预测精度更高,在轴力变化不同的四根桩上预测都很稳健,具有更强的泛化能力,在大型深水群桩基础的轴力预测中具有一定的工程应用价值。Accurately forecast for the axial force of deep-water group pile foundation pile is an important consideration on the safety and stability of foundation, due to the deep-water pile group foundation are influenced by the complex nonlinear relationship between the change of the pile axial force and environmental factors. In this paper, based on the prototype monitoring data of a bridge, the genetic algorithm optimized support vector machine model is introduced, the environmental factors influencing the variation of axial force of pile are deeply analysed, the multi factor prediction model of the pile axial force is built, and contrasts the predicted results by three models is cotrasted. Compared with the traditional SVM, RBF, the research results show that the GA-SVM model has higher accuracy of prediction and stronger generalization ability, and the prediction is very stable in the four different axial force of the pile. The GA-SVM model has certain engineering application value in the prediction of axial force in large deep-water group pile foundation.

关 键 词：深水群桩基础 支持向量机 遗传算法 轴力预测 

分 类 号：TU473.1[建筑科学—结构工程]