首页 | 本学科首页   官方微博 | 高级检索  
     

基于聚类匹配的煤层气压裂效果主控因素识别
引用本文:闵超,代博仁,石咏衡,杨兆中,李小刚,张馨慧. 基于聚类匹配的煤层气压裂效果主控因素识别[J]. 特种油气藏, 2022, 29(4): 135-141. DOI: 10.3969/j.issn.1006-6535.2022.04.019
作者姓名:闵超  代博仁  石咏衡  杨兆中  李小刚  张馨慧
作者单位:1.西南石油大学理学院,四川 成都 610500; 2.西南石油大学人工智能研究院,四川 成都 610500; 3.西南石油大学油气藏地质及开发工程国家重点实验室,四川 成都 610500; 4.国家管网集团油气调控中心,北京 100022
基金项目:国家科技重大专项“多层复杂煤体结构区煤储层直井压裂技术研究”(2016ZX05044-004-002);四川省科技计划项目“四川页岩气产业发展质量综合监测和评价技术研究与应用示范”(2020YFG0145);成都市科技局国际合作项目“基于深度学习的孔隙网络渗流模拟理论和技术探讨”(2020-GH02-00023-HZ)
摘    要:煤层气压裂效果与影响因素之间存在的非线性关系难以从机理层面进行系统分析,针对该问题,提出了一种基于聚类匹配的压裂效果主控因素识别方法。该方法从数据中挖掘影响因素的内在联系而非通过主观判断来分析压裂效果与影响因素之间的联系。首先,以压裂后的产气指标数据为研究对象,利用凝聚聚类方法对样本井进行分类和效果评价;其次,利用K-means聚类算法结合信息增益排序与相关性分析,对影响因素进行分类与筛选,从中选取前置液用量、携砂液用量、含气饱和度、含气量、垂直应力、支撑剂用量、破裂压力、加砂强度8个因素;最后,对筛选出的因素进行样本聚类,将聚类结果与压裂效果的评价分类结果进行聚类匹配,实现了压裂效果主控因素的识别。与其他主控因素识别方法对比,验证了该方法的有效性和可操作性。该研究可为优化二次压裂施工方案提供技术支持。

关 键 词:煤层气  压裂  凝聚聚类  主控因素  K-means聚类  信息增益
收稿时间:2021-06-17

Identification of Main Controlling Factors of Coalbed Methane Fracturing Effect Based on Cluster Matching
Min Chao,Dai Boren,Shi Yongheng,Yang Zhaozhong,Li Xiaogang,Zhang Xinhui. Identification of Main Controlling Factors of Coalbed Methane Fracturing Effect Based on Cluster Matching[J]. Special Oil & Gas Reservoirs, 2022, 29(4): 135-141. DOI: 10.3969/j.issn.1006-6535.2022.04.019
Authors:Min Chao  Dai Boren  Shi Yongheng  Yang Zhaozhong  Li Xiaogang  Zhang Xinhui
Affiliation:1. School of Science, Southwest Petroleum University, Chengdu, Sichuan 610500, China; 2. Institute for Artificial Intelligence, Southwest Petroleum University, Chengdu, Sichuan 610500, China; 3. State Key Laboratory of Oil and Gas Reservoir Geology and Exploration, Southwest Petroleum University, Chengdu, Sichuan 610500, China; 4. PipeChina Oil & Gas Control Center, Beijing 100022, China
Abstract:It is difficult to systematically analyze the nonlinear relationship between CBM fracturing effect and influencing factors from the perspective of mechanism, and a method based on cluster matching was proposed to identify the main controlling factors of fracturing effect. The method analyzed the connection between fracturing effect and influencing factors by tapping into the intrinsic connection of influencing factors based on data rather than by subjective judgment. Firstly, the data of gas production indicator after fracturing was taken as the object of study and the agglomerative clustering method was employed to classify and evaluate the effect of sample wells. Secondly, the influencing factors were classified and screened by K-means clustering algorithm in conjunction with information gain sequencing and correction analysis, from which 8 factors were selected: prepad fluid dosage, proppant-carrying fluid dosage, gas saturation, gas content, vertical stress, proppant dosage, fracturing pressure and proppant filling strength. Finally, the selected factors were clustered, and the clustering results were clustered and matched with the evaluation and classification results of fracturing effect, so as to identify the main controlling factors of fracturing effect. The effectiveness and operability of this method were verified by comparing it with other methods for identifying the main controlling factors. The study can provide technical support for optimizing the secondary fracturing plan.
Keywords:coalbed methane  fracturing  agglomeration clustering  main controlling factors  K-means clustering  information gain  
点击此处可从《特种油气藏》浏览原始摘要信息
点击此处可从《特种油气藏》下载全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号