利用测井资料进行裂缝的定量识别

The quantitative identification of fractures using well log data

1. 大庆石油学院, 安达 151400;2. 辽河油田钻采院, 盘锦 12401O

Daqing Petroleum Institute, Anda 151400

本文在分析砂泥岩剖面裂缝测并响应特点的基础上, 提出了一套由微球、双侧向、声波、密度和倾角洲共提取裂缝测井异常信息的方法, 然后根据松辽盆地古龙凹陷九口井65个裂缝发育层段资料, 分别利用综合判别函数和神经元网络方法分别建立了两个裂缝识别的数学模型, 其判推车分别是87％和93％, 表明神经元网络法识别裂缝效果较好。经处理实际资料说明上述裂缝定量识别方法是可行的。

Based on the characteristics of the we1l 1og data from the fractured sand and mud formation,we present a set of methods to extract the well log information of the fractures through the welllogs such as MSFL, DLL, DT, RHOB and HDT. Then, in accordance with 65 samples from 9wells in the Gulong depression of Songliao Basin, two mathematical models for the fractureidentification are set up by means of the comprehensive recognition function and the neura1network method respective1y. Their recognition accuracy rates are 87% and 93%, respectively.These show that the fracture identification effect by the neural network method is excellent. Theabove quantitative methods for fracture identification prove to be feasible via real data processing.