石油与天然气地质 ›› 2001, Vol. 22 ›› Issue (3): 249-252.doi: 10.11743/ogg20010313

• 油气空间 • 上一篇    下一篇

深度延迟人工神经网络判别水淹程度

唐文忠, 陈彬, 翟雨阳, 刘志远, 黄华   

  1. 中原油田分公司勘探开发科学研究院, 河南濮阳457001
  • 收稿日期:2011-05-31 出版日期:2001-09-25 发布日期:2012-01-16

APPLICATION OF DEPTH-DELAY ARTIFICAL NEURAL NETWORK TO DISCRIMINATION OF RESERVOIR FLOODED DEGREE

Tang Wenzhong, Chen Bin, Zhai Yuyang, Liu Zhiyuan, Huang Hua   

  1. Research Institute of Exploration and Development, Zhongyuan Oilfield Company, Puyang, Henan
  • Received:2011-05-31 Online:2001-09-25 Published:2012-01-16

PDF (PC)

818

被引次数

21

摘要：

深度延迟人工神经网络模型可以计算水淹层参数和判别水淹层,此方法避开了油层水淹后,混合液电阻率难以求准的问题,所建立测井信息与地层参数之间的非线性关系更符合油层水淹后的实际情况.深度延迟神经网络模型建立了储层地质参数与测井资料之间在深度上的动态关系,即一个深度点的地质参数由多个深度点的测井数据来描述,自动考虑了上下围岩的影响,而且测井信息与水淹层参数之间的复杂关系不需要具体的数学物理模型描述,只需有合适的样本集对网络训练即可获得解释模型.对东濮凹陷濮城油田沙河街组二段下亚段20口井的测井资料处理表明,用延迟人工神经网络模型计算水淹层剩余油饱和度,进而判别水淹程度,应用效果良好,符合率达82.4%.

关键词: 人工神经网络, 深度延迟, 测井资料, 剩余油饱和度, 水淹程度

Abstract:

Depth delay artificial neural network model is put up to calculate the parameters and the flooded degree of flooded reservoir.This method could avoid the problem of calculating the resistivity of mixed liquid after flooding,and the non linear relation between one depth point geo parameter and the log data of the present depth point and more depth points of the upper and lower beds is more accurate.To use this model,one have to select suitable sample assemblage to traim the network first,then the interpretation model is built.Use this model to process the log data obtained from twenty wells of the lower section of Sha 2 member in Pucheng Oilfield,Dongpu Depression,the pesult is good,the coincidence rate is up to 82.4%.

Key words: artificial neural network, depth delay, log data, remaining oil sauration, flooded degree

中图分类号: