BP神经网络在西湖凹陷烃源岩评价中的应用  

The Application of BP Neural Network to the Source Rocks Evaluation in Xihu Sag

作　　者：赵兴齐[1] 陈践发[1] 郭望[1] 刘高志[1] 陈斐然[1] 张文[2] 

机构地区：[1]中国石油大学(北京)油气资源与探测国家重点实验室,北京102249 [2]中国石油塔里木油田公司勘探开发研究院,新疆库尔勒841000

出　　处：《测井技术》2013年第5期567-571,共5页Well Logging Technology

基　　金：国家科技重大专项(No.2008ZX05023-1)大型油气田及煤层气开发资助

摘　　要：在测井评价中使用BP神经网络法拟合总有机碳含量,可以获得钻井剖面中连续、完整的烃源岩总有机碳含量变化和分布特征,解决了海上钻探井取样的困难。选取对总有机碳含量影响较大的常规测井曲线作为输入端,建立BP神经网络模型对西湖凹陷烃源岩总有机碳含量、生烃潜力指数和氢指数值进行拟合,从而为西湖凹陷烃源岩的评价提供了更加合理的地球化学参数。利用实测及BP神经网络拟合地球化学参数对西湖凹陷42口井的烃源岩有机质丰度综合分析认为,平湖组源岩有机质丰度高、厚度大,为凹陷主力烃源岩,其中平湖组中、上段为油气来源的主要层段;花港组源岩有机质丰度较低,对该区油气形成有一定贡献,为凹陷次要烃源岩。With BP （Back Propagation） neural network to fit TOC content in logging evaluation, we can get continuous and complete TOC variation and distribution of source rock in drilling section, which can overcome the limitation of sampling in offshore. Taking GR, CN, CAL, AC and R1Ld, affected by TOC significantly, as the input data, we build the BP neural network model, and simulate TOC content, S1＋S2 and IH, which supplies more reasonable geochemistry parameters to evaluate source rock in Xihu sag. After comprehensively analysis of organic matter abundance in 42 wells with geochemistry data simulated by actual measurements and BP neural network, it is concluded that Pinghu formation, particularly middle and upper section, is the major source rock, characterized by high abundance of organic matter and large thickness. By comparison, Huagang formation is the minor source rock, characterized by lower organic matter abundance, which makes a less contribution to the oil and gas reservoir.

关 键 词：测井解释 BP神经网络 总有机碳含量 烃源岩评价 西湖凹陷 

分 类 号：P631.84[天文地球—地质矿产勘探]