基于PSO-SVM模型的深基坑变形预测研究

Research on deformation prediction of deep foundation pit based on PSO-SVM model

深基坑变形预测一直是深基坑工程的一个重点研究课题,具有十分重要的理论意义和实际价值。支持向量机是一种基于结构风险最小化原理的机器学习算法,它具有很好的泛化能力,能够有效地解决小样本、非线性、高维数、局部极小等问题。本文将支持向量机(SVM)理论引入到深基坑的变形预测当中,同时,采用粒子群算法(PSO)来优化SVM的相关参数,将其预测结果与传统的支持向量机模型和BP神经网络模型的预测结果进行对比。结果表明,PSO-SVM模型用于变形预测是可行的。

The deformation prediction of deep foundation fit has been an important research topic with very important theoretical and practical significance.Support Vector Machine is a kind of machine learning algorithm based on structural risk minimization principle,which has good generalization ability to effectively address the small sample,nonlinear,high dimension and local minimum problems.This article introduces Support Vector Machine theory to predict the deformation of deep foundation fit and Particle Swarm Optimization（PSO） is used to optimize the SVM parameters.The predicted results are compared with that of conventional SVM and BP neural network.The results show that PSO-SVM model for deformation prediction is feasible.