基于浅层神经网络的准南煤田煤层气含量预测  

Prediction of Coalbed Methane Content in Zhunnan Coalfield Based on Shallow Neural Network

作　　者：胡永 罗聪 王博 胡振鹏 刘子强 仲劼 刘蒙蒙 HU Yong;LUO Cong;WANG Bo;HU Zhenpeng;LIU Ziqiang;ZHONG Jie;LIU Mengmeng(156 Coal Geological Exploration Team,Xinjiang Coal Geology Bureau,Urumqi 834022,China;School of Mining Engineering and Geology,Xinjiang Institute of Engineering,Urumqi 834023,China;Coalbed Methane Testing Institute of Xinjiang Uygur Autonomous Region,Urumqi 830009,China)

机构地区：[1]新疆煤田地质局一五六煤田地质勘探队,乌鲁木齐834022 [2]新疆工程学院矿业工程与地质学院,乌鲁木齐834023 [3]新疆维吾尔自治区煤炭煤层气测试研究所,乌鲁木齐830009

出　　处：《煤炭技术》2023年第3期138-142,共5页Coal Technology

摘　　要：煤层含气量是决定产能的重要因素,因此对含气量的预测是煤层气勘探领域的重点。利用煤层中含气量的不同导致煤层地球物理性质所产生的差异,优选出井深、自然伽马、电阻率、密度、中子孔隙度、声波时差等测井曲线,采用浅层神经网络对测井数据与含气量实验室检测数据开展研究,并与机器学习的研究结果进行了对比。研究发现,使用浅层神经网络法利用上述测井曲线可以对煤层含气量进行较为精准的预测,其相关系数可达0.91,准确度明显超过支持向量机、回归树等机器学习方法。研究结果对使用测井曲线准确预测煤层含气量,指导后续的煤层气开发工作具有较好的实际意义。Coalbed methane content is an important factor which determines the productivity, so the prediction of coalbed methane content is the focus of coalbed methane exploration. Based on the differences in geophysical properties of coal seams caused by different coalbed methane content in coal seams, the logging curves such as well depth, natural gamma ray, resistivity, density, neutron porosity and acoustic are optimized. The shallow neural network is used to study the logging data and coalbed methane content tested in the laboratory, and the results are compared with those of machine learning.It is found that the shallow neural network method can accurately predict the coalbed methane content of coal seam by using the above logging curves, and the correlation coefficient can reach 0.91. The accuracy is significantly higher than that of machine learning methods such as support vector machine and regression tree. The research results have good practical significance for accurately predicting coalbed methane content by logging curves and guiding subsequent coalbed methane development.

关 键 词：煤层气 测井 浅层神经网络 机器学习 

分 类 号：P631.81[天文地球—地质矿产勘探] P618.13[天文地球—地质学]