摘要
滑坡位移具有非线性特征,针对单变量的时间序列,首先引入相空间重构理论,将其扩展到多维的相空间中。再结合数据挖掘中的机器学习算法——支持向量回归算法(SVR)建立预测模型,并对李家湾滑坡的水平位移进行预测。试验结果表明,该模型具有非常高的精度(均控制在94%以上),可以充分的应用于滑坡灾害的预测和预报。
In view of the nonlinear characteristics of landslide displacement,the method based on phase space reconstruction can expand the single variable time series to multidimensional phase space.The horizontal displacement of Lijiawan Landslide is preformed for regression and prediction by phase space reconstruction and support vector regression.The test results show that this model is feasible with high precision(all controlled over 94%).Therefore,this method can be fully applied in landslide prediction and forecast.
出处
《长江大学学报（自科版）（上旬）》
2013年第2期76-78,90,共4页
JOURNAL OF YANGTZE UNIVERSITY (NATURAL SCIENCE EDITION) SCI ＆ ENG
基金
国家重点基础研究发展计划资助"973"计划(2011CB710605)
关键词
滑坡水平位移
相空间重构
支持向量回归
预测
landslide horizontal displacement
phase space reconstruction
support vector regression
prediction
