煤田声波测井曲线预测方法的对比研究  

Prediction Methods Comparison of Acoustic Logging Curve in Coal-field

作　　者：林庆西[1] 彭苏萍[1] 师素珍[1] 李娟[1] 汤尧[1] 喻梓靓[1] 

机构地区：[1]中国矿业大学北京煤炭资源与安全开采国家重点实验室,北京100083

出　　处：《煤矿开采》2014年第2期17-23,共7页Coal Mining Technology

基　　金：国家“973”项目(2009CB219603,2010CB226800,2009CB724601);国家自然基金重大项目(50490271,40672104);国家自然基金面上项目(40874071);煤炭联合基金项目(U1261203);国家“十二五”科技支撑计划(2012BAB13B01,2012BAC10B03)

摘　　要：为解决煤田地震勘探中声波测井资料不完整或缺失的问题,研究了3种声波测井曲线预测方法:经验公式法、地震属性分析法和神经网络法,并对这3种方法的原理和优缺点分别进行了对比与分析。以顾桥矿区为例,分别应用这3种方法对声波曲线进行了预测,并对预测结果进行了对比,结果表明:在煤田中,运用经验公式法、地震属性分析法、神经网络法所获得的声波曲线分辨率是逐步提高的。利用神经网络法可以获得最好的预测结果。In order to solve the problem of incomplete or missing data from acoustic logging in coal-field seismic exploration, 3 acoustic logging curve prediction methods including empirical formula, seismic attribute and neural network were researched. Their principles and relative merits were compared and analyzed. Taking Guqiao Colliery as an example, acoustic curve was predicted by applying 3 methods. Results showed that the acoustic curve resolution ratios were increased in sequence of empirical formula, seismic attribute and neural network in coal-field logging. Applying neural network could obtain the best prediction result.

关 键 词：煤田测井 声波测井曲线 经验公式 地震属性 神经网络 

分 类 号：TD315.63[矿业工程—矿井建设]