应用地震属性预测煤层顶板泥岩百分比含量分布  

Prediction of Mudstone Percentage Content for Coal Roof using Seismic Attributions

作　　者：孟召平[1,2] 郭彦省[2] 马辉[2] 

机构地区：[1]三峡大学三峡库区地质灾害教育部重点实验室,湖北宜昌443002 [2]中国矿业大学资源与安全工程学院,北京100083

出　　处：《中国矿业大学学报》2009年第2期274-279,共6页Journal of China University of Mining & Technology

基　　金：国家重点基础研究发展计划(973)项目(2007CB209405);国家自然科学基金项目(40772100);全国优秀博士学位论文作者专项资金项目(200247);高等学校博士学科点专项科研基金项目(20050290009);教育部留学回国人员科研启动基金项目

摘　　要：从地震属性分析入手,提出了用于煤层顶板泥岩百分比含量预测的地震属性分析方法和BP人工神经网络岩性预测方法.以淮南矿区潘东西四采区三维地震勘探区为依托,优选出平均瞬时相位、主频序列1,能量半衰时和主频斜率等4种地震属性作为13-1煤层顶板岩性预测分析的基本参数,结合已知钻孔资料,建立了煤层顶板泥岩百分比含量BP人工神经网络预测模型,运用训练好的网络对研究区13-1煤层顶板泥岩百分比含量进行了预测分析.结果表明,BP神经网络模型具有极强的非线性逼近能力,能真实反映煤层顶板岩性与地震属性之间的非线性关系,预测结果与实测值之间误差小,相对误差一般小于10%,地震属性可以用于煤层顶板岩性分布预测.Based on the analysis on seismic attributions, an analysis method of seismic attributes and prediction method of lithologic characters based on a BP artificial neural network were proposed for forecasting the mudstone percentage content of coal roof. Four usable seismic attributes, including average instantaneous phase, dominant frequency 1, energy half-time and spectral slope from peak to maximum frequency, were selected as the basic analysis parameters of prediction models of the roof lithologic character of 13-1 coal seam based on 3D seismic exploration area of mining section 4 in West Pandong of Huainan coal mining area. Combined with the real drill data, a BP artificial neural network prediction model of the mudstone percentage content of coal roof was established. Using a good training network model to predict and analyse roof lithologic character of 13-1 coal seam. The results show that the BP neural net- work has strong nonlinear approaching ability which can truly reflects the non-linear relationship between the lithologic characters of coal roof and seismic attributes. The relative error between predicted values and measured values is less than 10%, which indicate that the seismic attribute can be used in the distributing prediction of coal roof lithologic character.

关 键 词：地震属性 煤层顶板 泥岩百分比含量 人工神经网络 预测方法 

分 类 号：P631.4[天文地球—地质矿产勘探]