基于ABC-SVM的软基沉降预测研究

摘要

沉降预测一直是地基工程的一个重点研究项目.软土具有天然含水量高、塑性指数大、黏粒含量高等特殊的工程特性,因此,建立在软土上的地基沉降难以预测,对于工程建设有一定的隐患.支持向量机在解决样本数量小、非线性问题上有其特有的优势.利用人工蜂群算法对支持向量机的参数进行优化后建立SVM模型,对木兰溪防洪工程的沉降问题进行预测,将其预测结果与传统的支持向量机模型以及曲线拟合预测方法结果进行对比.最终证明了ABC-SVM模型在软基沉降预测上的可用性.%The prediction of the settlement is always an important research topic in foundation engineering. Soft soil has engineering features of high moisture content,high plastic exponent and high clay content,so it is hard to predict the settlement of soft soil foundation which is dangerous to the engineering construction. The support vector machines are good at dealing with problems that are nonlinear and have little sample size. We use artificial bee colony to optimize the selection of support vector machines'parameters to build SVM model. And the model is used to predict the settlement of flood control project in Mulan stream. The predicting outcomes are used to comparing with the traditional support vector machines and the curve-fitting method's outcomes. It turns out the usability of the ABC-SVM in the prediction of soft soil foundation.