基于支持向量机的时深建模方法TIME-DEPTH MODELING METHOD BASED ON SVM
袁航,谢玮,毕臣臣,岳占伟,刘伟,刘学清
YUAN Hang,XIE Wei,BI Chenchen,YUE Zhanwei,LIU Wei,LIU Xueqing
摘要(Abstract)：
为了解决海上勘探初期地震资料解释过程中的时深关系问题,提出基于粒子群和支持向量机的时深建模方法,利用粒子群算法来优化支持向量机的参数。首先由合成地震记录标定得到每口井的时深关系;然后利用支持向量机方法建立时间与深度之间的关系模型;最后以此模型对时间域等T0图进行转换,从而得到深度域的构造图。实际测井数据和三维地震资料试验结果表明,该方法建立的时深关系模型适用于整个研究区,且时间域转换深度与井分层深度、构造深度之间的误差较小,能够满足精细构造解释的精度要求。
In order to solve the problem of time-depth relationship during the seismic data interpretation for the initial offshore exploration,the time-depth modeling method was proposed based on the particle swarm and support vector machines( SVM),and moreover the SVM parameters were optimized by Particle Swarm Optimization( PSO). Firstly the time-depth relationships for each well were obtained by the synthetic seismogram calibration;and then the relationship model between the time and depth was established with the help of SVM method; finally with the help of the model,the conversion was conducted for the T0 contour in the time domain,then the structural map for the depth domain was obtained. The test results of the actual logging data and 3 D seismic data show that the time-depth relationship model built by the method is suitable to the whole study area,and moreover the errors are pretty small between the inverted depth in the time domain and the well separated depth and the structural depth. In a word,the model can meet the precision requirements of the fine structure interpretation.
关键词(KeyWords)：
地震资料解释;粒子群算法;支持向量机;时深关系建模
seismic data interpretation;PSO;SVM;time-depth relationship modeling
基金项目(Foundation): 国家科技重大专项“前陆冲断带及复杂构造地震成像关键技术与构造圈闭刻画”(2016ZX05003-003);; 中国石油化工集团公司项目“基于绕射波成像的缝洞储集体预测技术研究”(JP16033)联合资助
作者(Author):
袁航,谢玮,毕臣臣,岳占伟,刘伟,刘学清
YUAN Hang,XIE Wei,BI Chenchen,YUE Zhanwei,LIU Wei,LIU Xueqing
DOI: 10.19597/J.ISSN.1000-3754.201801111
参考文献(References)：
- [1]凌云,郭建明,郭向宇,等.油藏描述中的井震时深转换技术研究[J].石油物探,2011,50(1):1-13.
- [2]张志明,曹丹平,印兴耀,等.时深转换中的井震联合速度建模方法研究与应用现状[J].地球物理学进展,2016,31(5):2276-2284.
- [3]郭璃,邓勇,赵顺兰,等.特殊地质条件下的时深转换方法探讨:以涠西南凹陷南部斜坡带X井为例[J].地球物理学进展,2017,32(1):152-159.
- [4]李伍志,王璞珺,张功成,等.珠江口盆地深部基底地层的地震时深转换研究[J].地球物理学报,2011,54(2):449-456.
- [5]张英德,彭佳勇,郝立业,等.海外深水复杂地质条件下时深转换难点及技术对策[J].地球物理学进展,2012,27(4):1484-1492.
- [6]杨海长,赵志刚,张生强,等.拟合公式法时深转换在深层应用中的探讨[J].地球物理学进展,2016,31(4):1852-1856.
- [7]张华军,肖富森,刘定锦,等.地质构造约束层速度模型在时深转换中的应用[J].石油物探,2003,42(4):521-525.
- [8]杨德兴.库车坳陷地震速度建场及变速成图方法研究[D].北京:中国石油大学,2008.
- [9]徐立恒,鲜波,薛玉英,等.高精度地震时深转换方法研究及应用[J].吉林大学学报(地球科学版),2014,44(5):1712-1719.
- [10]Vapnik V N.The nature of statistical learning theory[M].New York:Sptinger Verlag,1995.
- [11]于代国,孙建孟,王焕增,等.测井识别岩性新方法——支持向量机方法[J].大庆石油地质与开发,2005,24(5):93-95.
- [12]Eberhart R,Kennedy J.A new optimizer using particle swarm theory[C]//Proceeding of the IEEE 6th International Symposium on Micro Machine and Human Science,1995.
- [13]张庆,刘丙杰.基于PSO和分组训练的SVM参数快速优化方法[J].科学技术与工程,2008,8(16):4613-4616.
- [14]谢玮,刘斌,钱艳苓,等.基于粒子群优化最小二乘支持向量机的裂缝及缝洞充填物识别[J].大庆石油地质与开发,2017,36(2):135-142.