v-SVC算法在地震与爆破识别及窗长度选取中的应用

英文篇名：v—SVC algorithm applied in earthquake and explosion recognition and the choice of window length
中文刊名：地震地磁观测与研究
英文刊名：Seismological and Geomagnetic Observation and Research
作者：黄汉明 ; 边银菊 ; 卢世军 ; 李锐 ; 蒋正锋
英文作者：Huang Hanming~(1)) ; Bian Yinju~(2)) ; Lu Shijun~(1)) ; Li Rui~(1)) and Jiang Zhengfeng~(1)) 1) College of Computer and Information Engineering ; Guangxi Normal University ; Guilin 541004 ; China 2) Institute of Geophysics ; China Earthquake Administration ; Beijing 100081 ; China
中文关键词：支持向量机 ; 分类识别 ; 窗长度选取 ; 人工爆破 ; 香农熵
英文关键词：support vector machine ; classification and recognition ; choice of window length ; explosion ; shannon entropy
出版日期：2010-06-25
机构：广西师范大学计算机科学与信息工程学院;中国地震局地球物理研究所;
年：2010
期：03
出版单位：地震地磁观测与研究

摘要

对天然地震与人工爆破的波形记录,本文用v—SVC支持向量分类机对由波形记录获取的香农熵特征进行了分类识别,效果较好;并对波形记录选取不同的信号窗长度,用v—SVC支持向量分类机分别进行了识别检验。结果表明:窗长度对识别效果有影响,以窗长度为2000点的识别效果最好,识别率达98%。这也表明,在地震与爆破的识别中,合理地选取波形记录的信号窗长度也是重要的。
In this paper, the v—SVC algorithm is applied in the classification of earthquake and explosion basing upon Shannon entropy features extracted from seismic wave records. The recognition effect is approving. Several different window lengths of wave record are used for feature extraction. Classifications by v—SVC are carried out for recognition tests. The results show that window length is also an important factor for recognition rate. The best window length is 2000 sampling points which achieves 98% recognition rate. This means that appropriate window length of seismic signal may also be an important role in the classification of earthquake and explosion.

引文

边银菊.遗传BP网络在地震和爆破识别中的应用[J].地震学报,2002,24(5) :516-524．
    高成,高志,余啸海.Matlab小波分析与应用-第2版[M].北京:国防工业出版社,2007．
    郭丽娟,孙世宇,段修生.支持向量机及核函数研究[J].科学技术与工程,2008,8(2) :487-490．
    和雪松,李世愚,沈萍,冯全雄.用小波包识别地震和矿震[J].中国地震,2006,22(4) :425-434．
    黄汉明,边银菊,卢世军,蒋正锋,李锐.天然地震与人工爆破的波形小波特征研究[J].地震学报(待发表),2010．
    刘希强,沈萍,张玲,李永红.用小波变换能量线性度方法识别天然地震与爆破或塌方[J].西北地震学报,2003,25(3) :204-209．
    王睿.关于支持向量机参数选择方法分析[J].重庆师范大学学报(自然科学版),2007,24(2) :36-38,42．
    杨选辉,沈萍,刘希强,郑治真.地震与核爆识别的小波包分量比方法[J].地球物理学报,2005,48(1) :148-156．
    Vapnik V.Statistical Learning Theory[M].Wiley,NY,1998．
    Pai-Hsuen Chen,Chih-Jen Lin,and Bernhard Scholkopf.A Tutorial on v-Support Vector Machines[J].Applied Stochastic Models in Business and Industry,2005,2(2) :111-136．