摘要
烃源岩评价是含油气盆地油气资源评价的重要基础,但烃源岩样品的地化测试受样品来源、分布及测试费用等因素的共同限制,难以满足勘探的需求。系统阐述了烃源岩总有机碳含量(TOC)的测井定量评价方法,并对X凹陷P层组的一口取心井的TOC数据进行分析。首先描述了单因素法、PCA法、多元回归分析法以及神经网络法计算TOC含量的优缺点,其次结合研究区的取心样品数据以及研究区的地质特点,对4种模型计算的效果进行分析,TOC计算结果表明:神经网络模型效果最好,其次是PCA模型,再次是多元回归模型和单因素模型;计算平均均方根误差分别为2.08%,14.59%和15.23%。最终确定神经网络模型对研究区烃源岩TOC计算的效果更好。
Identification and evaluation of source rock is on the basis of geological study for hydrocarbon.But geochemical testing of source rock samples is limited by the source,distribution and testing cost of samples.It is difficult to meet the requirements of exploration.This paper systematically expounds the logging quantitative evaluation method of TOC in source rocks.It also analyses the calculation results in a coring well of P formation in X sag.Firstly,the advantages and disadvantages of single factor method,PCA method,multiple regression analysis method and neural network method for calculating TOC content are described.Secondly,it is combined with the core sample data of the study area and the geological characteristics of the study area.The calculation results of the four models are analyzed.The results show that the neural network model has the best effect,and the average root mean square error is 2.08%,followed by PCA model.The average root mean square errors are 14.59%and 15.23%respectively.It is proved that the TOC content calculated by the neural network model is more effective.
作者
刘继龙
LIU Jilong(College of GeoSciences,Northeast Petroleum University,Daqing,Heilongjiang 163318,China;State Key Laboratory of Unconventional Oil and Gas Accumulation and DevelopmentEstablished by Provincial and Ministry Departments,Daqing,Heilongjiang 163318,China)
出处
《中国锰业》
2019年第3期54-58,共5页
China Manganese Industry
关键词
烃源岩
测井定量评价
多元回归法
神经网络法
Source rock
Logging qualitative identification
Multiple regression method
Neural network method
