自动识别碳酸盐岩储层裂缝的方法

A METHOD OF AUTOMATIC IDENTIFYING CARBONATE RESERVOIR FRACTURES

四川地区大部分油气田的储层属于碳酸盐岩地层，其特点是非常致密的，孔隙度小、渗透率低。在碳酸盐岩中油气的储集空间为孔隙和裂缝，而油气的运移通道常常以裂缝为主。所以在碳酸盐岩地层中，对裂缝的识别与评价方法的研究具有重要意义。目前对裂缝研究的主要方法还是人工分析对比法，不仅工作量大，而且还会由于种种人为因素的影响，使其解释质量下降。这里试图把计算机的模式识别技术引入这个领域，对测井资料作自动解释。在测井解释中，已有人把此技术用来识别油气水层，取得了一定的效果。但将测井信息与计算机的模式识别技术相结合来研究裂缝特征的工作在国内外还较少。笔者采用模式识别中一种较新的方法──应用证据推理的树分类器法，对单井储层裂缝进行综合识别，这种方法的特点兼有分类器和证据推理的优点，理论上已证明该分类器是贝叶斯（Ｂａｙｅｓ）最小错误概率的分类器。利用此方法在川东卧龙河气田和川西南矿区进行试验，取得了满意效果。

The carbonate rocks with good compactability, small porosity and low permeability are the reservoirs of greater part oil-gas fields in Sichuan Basin. The storage spaces of oil-gas in carbonate reservoirs are pores and fractures, but the main migration pathways are often the factures. So it is of important significance to identity the fractures and research the evaluation methods. The manual analysis comparision method currently used is not only troublesome but also influences the interpretive quality. Computer modelling identification technique is introduced to automatically interpret logging data. A new method of the modelling technique, i. e. tree classifier method, is used to comprehensively indentify the fractures of individual well reservoirs in Wolonghe gas field in East Sichuan and in Southwest Sichuan, and satisfactory results are obtained. The method has the advantages of tree classifier and evidentiary inference. The classifier is proved by theory to be the one with Bayes smallest probability of error.