基于集成聚类均质度与储层地质边界度联合的测井曲线自动分层

Integration of ensemble clustering homogeneity and geological marginal degree based automatic layering with logging curves

现有的储层分层方法主要对焦于地质边缘奇异点的理解,而忽略了储层相应曲线集隐含的均质性信息.为此,开展了基于集成聚类的均质度与储层地质边界度联合测井曲线自动分层方法研究.具体地,通过利用集成聚类联合共生关联矩阵进行均质地层信息的编码,提取曲线均质度特征,并利用简单的阈值分割实现对均质地质区域的初始分割;同时,为对区域信息形成补充,提出了一种相关差特征获取曲线集各点的地质边界度.通过联合应用测井曲线集蕴含的均质区域信息与异构(奇异)地质边缘信息实现了一种灵活利用测井曲线集进行自动分层的方法.取用自然伽马、井径等7条常规测井曲线,对大庆油田陆相坳陷盆地齐家凹陷工区多口井进行实验与应用,研究结果表明,相对于传统分层方法更加准确高效,精度更高,特别在工区内的薄层和薄互层上取得较好的应用效果,不仅能高效稳定的实现单砂体级别的储层目标自动精细分层,还在一定程度上增加了对人工解译结果的补充,具有一定的实用性,可为提高测井储层解释的准确性提供较可靠的分层依据.

The existing reservoir division methods are only suitable for conventional rock formation division, and cannot be used for the fine description of unconventional oil and gas resources such as medium-shallow tight oil and thin sand.Therefore, an automatic stratification method based on the integrated clustering of the combined logging curve of the homogeneity and the geological boundary of the reservoir is proposed to solve the structural division of tight sandstone reservoirs.Specifically, by using the integrated clustering information and the co-occurrence correlation matrix to encode the homogeneous formation information of the logging data, the homogeneity feature of the logging curve is extracted, and a simple threshold segmentation is used to realize the preliminary division of the homogeneous geological area.Meanwhile, a novel correlation difference feature is designed to calculate the geological marginal degree. Finally, by integrating the obtained homogeneous regional information and the geological marginal degree feature, the proposed automatic layering methods can be realized. Seven conventional logging curves, such as conventional natural gamma and well diameter, are selected as input parameters.Experiments and applications have been carried out on the actual logging data of multiple medium-shallow layers of tight oil in the Qijia sag work area of the continental depression basin of Daqing Oilfield. The research results show that this method is more accurate and efficient than the traditional layering method, and has higher precision,especially for thin layers and thin interbedded layers in the work area. The automatic fine stratification of layer targets also supplements the results of manual interpretation to a certain extent, indicating that this method has certain practicability and can provide a reliable stratification basis for improving the accuracy of logging reservoir interpretation.