基于混沌-支持向量机的边坡位移预测  

Chaotic-SVM Based Displacement Prediction of Side Slope

作　　者：曹延飞[1] 李源[1] 贾磊[1] 

机构地区：[1]石家庄经济学院,河北石家庄050031

出　　处：《石家庄铁路职业技术学院学报》2014年第1期79-85,共7页Journal of Shijiazhuang Institute of Railway Technology

基　　金：河北省教育厅高等学校科学技术研究项目(编号:Z2011224)

摘　　要：通过平均互信息法确定延迟时间、虚假邻点法确定嵌入维数,从而实现混沌时间序列相空间的重构,揭示边坡地质体的变形规律;并以这些数据作为支持向量机的训练样本,用支持向量机的回归特性预测边坡的变形,把由于多变量耦合关系导致的边坡体位移时间序列,利用单变量方法进行分析。结果表明Chaotic-SVM模型具有拟合精度高、泛化能力强的特点,在岩土工程非线性预测中具有一定的应用价值。This study realized the phase space reconstruction of chaotic time series and revealed the deforming law of slope by using average mutual information to determine the delayed time and false nearest neighbor methods to determine the embedded dimension. The data was used as the training sample of the support vector machine. Displacement of side slope is predicted by the regression feature of the support vector machine. Time series of the side slope displacement caused by multivariate coupling was analyzed by univariate approach. The result shows that Chaotic-SVM model has the characteristics of high precision and strong generalization ability, which can be applied in the nonlinear prediction of geotechnical engineering.

关 键 词：位移混沌动力模型 相空间重构 边坡安全监控 变形预测 

分 类 号：O31[理学—一般力学与力学基础]