基于ε-SVR的滑坡位移预测  

PREDICTION OF LANDSLIDE DISPLACEMENT BASED ON ε-SUPPORT VECTOR REGRESSION MACHINE

作　　者：方臣[1] 王正海[1,2] 耿欣[1] 段建军[1] 

机构地区：[1]中山大学地球科学系,广东广州510275 [2]广东省地质过程与矿产资源探查重点实验室,广东广州510275

出　　处：《防灾减灾学报》2012年第2期39-43,共5页Journal of Disaster Prevention And Reduction

基　　金：教育部博士点基金项目(20090171120018);教育部高校基本科研业务费中山大学优秀青年教师培育计划项目(2010-32110-3161319)

摘　　要：我国是一个地质灾害多发的国家,特别是滑坡发生的次数比较多、危害性比较大。因此对滑坡的位移进行监控预测有着十分重要的意义。对于滑坡位移变化的非线性问题,可以利用支持向量机在回归分析中的方法——ε-支持向量回归机(ε-SVR)进行预测,该方法基于统计学理论,在处理小样本、非线性、高维数等问题上有一定的优势。以福建八尺门滑坡的监测数据为例,将前面的17个位移数据作为学习样本,后面的6个位移数据作为预测样本,采用不同的核函数分别进行位移预测来与原始监测值进行对比,比较其预测精度。结果显示,该方法产生的预测值与原始监测值之间的误差比较小,其位移变化趋势与原始数据的变化趋势也基本一致,这说明该方法预测精度高,实用性强。Our country is a geological disaster-prone countries, especially the landslide happened more frequently with more harmfulness. So the prediction of landslide displacement has a very important meaning. To solve the nonlinear problem about forecast the change of landslide displacement, we can use a new method--a-support vector regression machine that come from support vector machine. This method is based on Statistical learning theory, which has many advantages to deal with small sample, nonlinear, high dimension problem. Taking the displacement data at Bachimen of Fujian as an example--the former 17 displacement data as study samples, the last of the 6 displacement data as prediction of samples, we use different kernel functions on displacement prediction respectively to compare with the original monitoring values and the prediction accuracy. The results show that error between the prediction value and the original monitoring value is relatively small. And the displacement change trend is also consistent. It shows that the prediction precision and function of this method is good.

关 键 词：滑坡 ε-支持向量回归机 核函数 位移预测 

分 类 号：P642.22[天文地球—工程地质学]