基于测井资料的BP神经网络的煤体结构预测  

Prediction of coal structure with BP neural network based on shaft logging data

作　　者：刘振明[1] 王延斌[1] 韩文龙[1] 倪冬[1] 张崇瑞 Liu Zhenming;Wang Yanbin;Han Wenlong;Ni Dong;Zhang Chongrui(School of Geoscience and Surveying Engineering, China University of Mining and Technology, Beijing, Haidian, Beijing 100083, China)

机构地区：[1]中国矿业大学(北京)地球科学与测绘工程学院,北京市海淀区100083

出　　处：《中国煤炭》2018年第6期38-41,55,共5页China Coal

基　　金：国家科技重大专项资助项目(2017ZX05064-003-001)

摘　　要：基于柿庄南3号煤层测井和钻井岩芯数据,结合地质强度因子对煤岩煤体结构进行定量表征,通过因子分析完成对声波时差、体积密度、自然伽马、井径、补偿中子以及深层向电阻率测井曲线的优选,并将其结果作为模型输入参数,利用BP神经网络方法,建立了该地区的煤层煤体结构GSI值的预测模型。预测数值与目标数值具有高度的吻合度,并再次利用多元线性回归方法进行对比。结果表明,BP神经网络方法具有更好的适用性,为以后煤体结构预测模型的建立提供了新思路。Based on the logging and drilling rock core data of No.3 coal seam in South Shizhuang,combined with geological intensity factor to describe coal rock structure,applying factor analysis to optimize shaft logging curves of acoustic jet lag,volume density,natural gamma,shaft diameter,compensation neutron and deep direction resistivity,and taking the results as model input parameters and adopting BP neural network to build prediction model of GSI value of coal structure in this area.The predicted values were highly consistent with the target values,and were compared through multiple linear regression methods.The result showed that the predicted values was more applicable and provided a new way to construct prediction model in the future.

关 键 词：煤体结构 测井曲线 BP神经网络 因子分析 

分 类 号：P585.2[天文地球—岩石学]