利用测井曲线定量识别安泽区块煤体结构

Well logs applied to quantitatively identify coal structure in the Anze area

以沁水盆地南部安泽区块山西组3号煤岩为研究对象,利用11口评价井的钻井取心和常规测井曲线资料,利用BP神经网络方法,对煤体结构进行了定量识别.研究结果表明:根据宏观观察的取心破碎情况,安泽区块山西组3号煤岩煤体结构可划分为原生结构煤、碎裂结构煤、碎粒结构煤和糜棱结构煤四类;随着破碎程度的增强主要表现为自然伽马、声波时差、补偿中子和井径的增大,密度和深电阻率降低;不同煤体结构测井曲线数值存在大量重叠,利用测井曲线识别煤体结构具有多解性;选取自然伽马、补偿中子、井径、声波时差、密度和深电阻率6条测井曲线,利用BP神经网络方法,实现了基于常规测井曲线和BP神经网络的煤体结构定量识别.该方法可有效提高煤体结构识别准确性.

This study took the No.3 coal seam of Shanxi Formation in Anze area,Qinshui basin as an example,and identified different coal structures quantitatively using BP neural network method by comprehensive utilization of cores and conventional logging curve data of 11 coring wells.The results show that according to the macroscopic fragmentized degree of No.3 coal seam in Anze area,coal structure can be divided into undeformed,cataclastic,granulated and mylonitic coal.With the degree of fragmentation increasing,it commonly performance as the increasing of gamma ray,acoustic time,compensation neutron and hole diameter,while the decreasing of density and resistivity.The single log response characteristics of different coal structure is obscure and insensitive,and overlaps exist between different coal structures.Gamma ray,compensation neutron,hole diameter,acoustic time,density and deep resistivity logging curve are selected.By using BP neural network method,the quantitative identification of coal structure has been realized.This method can improve the accuracy of identification of coal structure effectively.