基于正演模拟和SVM的瓦斯突出危险区预测
|
摘要
以瓦斯地质基本理论为基础,利用地质和钻井数据建立了含瓦斯煤层的地质和地球物理模型.对所建立的地球物理模型,通过有限差分正演模拟方法获得了正演地震剖面.通过对地震剖面煤层反射波的属性分析,获得了相应的地震属性,在此基础上,运用支持向量机(SVM)方法对瓦斯突出危险区进行了预测.结果表明:运用惩罚参数C=32,γ=78.125×10-4的RBF核函数和所建模型对测试样本钻孔数据进行分类预测,预测精度为80%;对随机选择的训练样本数据进行回代预测,预测精度达到90%,为利用叠后地震数据预测瓦斯突出危险区提供了一条新途径.
Based on coal gas theory,geological and geophysical models of gaseous coal seam were built by geological and drilling data.For the established models,forward modeling seismic sections were achieved by the computation of finite difference.By analysis of the coal reflection attributes of those modeling sections,seismic attributes were obtained.The SVM(Support Vector Machine) method was also used in our work to predict gas outburst risk area based on the extracted seismic attributes.The results show reveals that: firstly,the prediction accuracy is 80% in classifying and predicting drilling data of testing samples by using RBF kernel function with parameters C=32 and γ=78.125×10-4 and the established model;secondly,the prediction accuracy is as high as 90% when the prediction of randomly-selected training sample data is carried out recursively.This has provided a new approach to predict gas outburst risk area by using post-stack seismic data.
Based on coal gas theory,geological and geophysical models of gaseous coal seam were built by geological and drilling data.For the established models,forward modeling seismic sections were achieved by the computation of finite difference.By analysis of the coal reflection attributes of those modeling sections,seismic attributes were obtained.The SVM(Support Vector Machine) method was also used in our work to predict gas outburst risk area based on the extracted seismic attributes.The results show reveals that: firstly,the prediction accuracy is 80% in classifying and predicting drilling data of testing samples by using RBF kernel function with parameters C=32 and γ=78.125×10-4 and the established model;secondly,the prediction accuracy is as high as 90% when the prediction of randomly-selected training sample data is carried out recursively.This has provided a new approach to predict gas outburst risk area by using post-stack seismic data.
引文
[1]何继善.瓦斯突出地球物理研究[M].长沙:中南工业大学出版社,1999:1-28.
[2]BACKUS G E.Long-wave elastic anisotropy pro-duced by horizontal layering[J].Journal of Geophys-ics Research,1962,67(11):4427-4440.
[3]POSTMA G W.Wave propagation in a stratified me-dium[J].Geophysics,1955,20(4):780-806.
[4]HUDSON J A.Wave speeds and attenuation of elas-tic waves in material containing cracks[J].Geophys-ics Journal International,1981,64(1):133-150.
[5]HUDSON J A.Seismic wave propagation throughmaterial containing partially saturated cracks[J].Geophysics Journal International,1988,92(1):33-37.
[6]HUDSON J A,LIU E,CRAMPIN S.The mechani-cal properties of materials with interconnected cracksand pores[J].Geophysics Journal International,1996,124(1):105-112.
[7]LIU E,HUDSON J A,POINTER T.Equivalentmedium representation of fractured rock[J].Journalof Geophysics Research,2000,105:2981-3000.
[8]陈同俊,崔若飞,刘恩儒.VTI型构造煤AVO正演模拟[J].煤炭学报,2009,34(4):438-442.CHEN Tong-jun,CUI Ruo-fei,LIU En-ru.AVOforward modeling for VTI coal[J].Journal of ChinaCoal Socity,2009,34(4):438-442.
[9]陈同俊,王新,崔若飞.基于方位AVO正演的HTI构造煤裂隙可探测性分析[J].煤炭学报,2010,35(4):640-644.CHEN Tong-jun,WANG Xin,CUI Ruo-fei.Thedetectability analysis on HTI tectonic coal cracks byazimuthal AVO's forward modeling[J].Journal ofChina Coal Socity,2010,35(4):640-644.
[10]张学工.关于统计学习理论与支持向量机[J].自动化学报,2000,26(1):32-42.ZHANG Xue-gong.Introduction to statistical learn-ing theory and support vector machines[J].ActaAutomatica Sinica,2000,26(1):32-42.
[11]VAPNIK V N.Statistical learning theory[M].New York:Springer,1998:732.
[12]彭苏萍,高云峰,杨瑞召,等.AVO探测煤层瓦斯富集的理论探讨和初步实践:以淮南煤田为例[J].地球物理学报,2005,48(6):1475-1486.PENG Su-ping,GAO Yun-feng,YANG Rui-zhao,etal.Theory and application of AVO for detection ofcoalbed methane:A case from the Huainan coalfield[J].Chinese Journal of Geophysics,2005,48(6):1475-1486.
[13]姚宏善.基于支持向量机的财务困境预测研究[D].武汉:华中科技大学控制科学与工程系,2006:79-80.
[14]HSU C,LIN C.A comparison of methods for mul-ticlass support vector machines[J].IEEE Transac-tions on Neural Networks,2002,13(2):415-425.
[15]国家煤矿安全监察局.防治煤与瓦斯突出规定[M].北京:煤炭工业出版社,2009:29.
[2]BACKUS G E.Long-wave elastic anisotropy pro-duced by horizontal layering[J].Journal of Geophys-ics Research,1962,67(11):4427-4440.
[3]POSTMA G W.Wave propagation in a stratified me-dium[J].Geophysics,1955,20(4):780-806.
[4]HUDSON J A.Wave speeds and attenuation of elas-tic waves in material containing cracks[J].Geophys-ics Journal International,1981,64(1):133-150.
[5]HUDSON J A.Seismic wave propagation throughmaterial containing partially saturated cracks[J].Geophysics Journal International,1988,92(1):33-37.
[6]HUDSON J A,LIU E,CRAMPIN S.The mechani-cal properties of materials with interconnected cracksand pores[J].Geophysics Journal International,1996,124(1):105-112.
[7]LIU E,HUDSON J A,POINTER T.Equivalentmedium representation of fractured rock[J].Journalof Geophysics Research,2000,105:2981-3000.
[8]陈同俊,崔若飞,刘恩儒.VTI型构造煤AVO正演模拟[J].煤炭学报,2009,34(4):438-442.CHEN Tong-jun,CUI Ruo-fei,LIU En-ru.AVOforward modeling for VTI coal[J].Journal of ChinaCoal Socity,2009,34(4):438-442.
[9]陈同俊,王新,崔若飞.基于方位AVO正演的HTI构造煤裂隙可探测性分析[J].煤炭学报,2010,35(4):640-644.CHEN Tong-jun,WANG Xin,CUI Ruo-fei.Thedetectability analysis on HTI tectonic coal cracks byazimuthal AVO's forward modeling[J].Journal ofChina Coal Socity,2010,35(4):640-644.
[10]张学工.关于统计学习理论与支持向量机[J].自动化学报,2000,26(1):32-42.ZHANG Xue-gong.Introduction to statistical learn-ing theory and support vector machines[J].ActaAutomatica Sinica,2000,26(1):32-42.
[11]VAPNIK V N.Statistical learning theory[M].New York:Springer,1998:732.
[12]彭苏萍,高云峰,杨瑞召,等.AVO探测煤层瓦斯富集的理论探讨和初步实践:以淮南煤田为例[J].地球物理学报,2005,48(6):1475-1486.PENG Su-ping,GAO Yun-feng,YANG Rui-zhao,etal.Theory and application of AVO for detection ofcoalbed methane:A case from the Huainan coalfield[J].Chinese Journal of Geophysics,2005,48(6):1475-1486.
[13]姚宏善.基于支持向量机的财务困境预测研究[D].武汉:华中科技大学控制科学与工程系,2006:79-80.
[14]HSU C,LIN C.A comparison of methods for mul-ticlass support vector machines[J].IEEE Transac-tions on Neural Networks,2002,13(2):415-425.
[15]国家煤矿安全监察局.防治煤与瓦斯突出规定[M].北京:煤炭工业出版社,2009:29.
版权所有：© 2023 中国地质图书馆 中国地质调查局地学文献中心