基于支持向量机的多维地震时间序列建模  

THE MODELING OF MULTI-DIMENSION SEISMIC TIME SERIES BASED ON THE SUPPORT VECTOR MACHINE

作　　者：武安绪[1] 李平安[1] 鲁亚军[1] 穆会泳[1] 苏小非[1] 刘学谦 

机构地区：[1]北京市地震局,北京100080 [2]河南省地震局,河南郑州450016

出　　处：《东北地震研究》2006年第4期30-34,共5页Seismological Research of Northeast China

基　　金：地震科学联合基金项(A07058)

摘　　要：文中首先介绍了支持向量机回归方法,然后对其进行了数值仿真实验,最后利用该方法对首都圈地区的地震活动进行回归建模与内符检验。分析认为,利用该方法可以获得较强的多维地震样本学习、建模与外推泛化能力,具有较高的中长期地震危险性估计能力。Firstly, the regression method of support vector machine is introduced in this paper. Secondly, effective experiment is done by using of the emulation data. Lastly, regression model of seismic activity is established in Capital Circle. By contrasting and analyzing, we think that the result is fairly effective. It has shown that support vector machine method have ability of multi-dimension learning and forecast, and may be a new effective processing method in analyzing of deformation data.

关 键 词：支持向量机 地震时间序列 回归 预测 

分 类 号：P315.7[天文地球—地震学]