基于粒子群优化最小二乘支持向量机的裂缝及缝洞充填物识别PSO-LSSVM-BASED IDENTIFYING METHOD FOR THE FRACTURE-VUG FILLINGS
谢玮,刘斌,钱艳苓,孙炜,史飞洲,李玉
XIE Wei,LIU Bin,QIAN Yanling,SUN Wei,SHI Feizhou,LI Yu
摘要(Abstract)：
碳酸盐岩储层中发育的缝洞是油气的主要储集空间和渗流通道,缝洞充填物影响着储层的渗流能力和储集能力,碳酸盐岩储层中裂缝及缝洞充填物的识别对于此类储层的勘探开发具有重要意义。基于常规测井资料对裂缝和缝洞充填物的响应特性,提出综合应用粒子群算法(PSO)和最小二乘支持向量机(LSSVM)的裂缝及缝洞充填物识别方法 (PSO-LSSVM),并将该方法和BP神经网络分别应用于滨里海盆地东缘石炭系碳酸盐岩储层裂缝及缝洞充填物的识别。对比分析两种方法的识别结果,PSO-LSSVM的识别效果比BP神经网络好。利用PSO-LSSVM方法得到的识别结果与FMI电成像测井图像及岩心资料得到的结果有较好的一致性。
Fracture-vugs well-developed in carbonate reservoirs are the main part of the accumulating space and seepage channels for the oil and gas,but the fillings in the fracture-vugs influence the storage and seepage capacities of the reservoirs,so the identifications of the fractures and fillings play an important roles in the exploration and development of this kind of the carbonate reservoir. Based on the response characteristic of the conventional well logging data to the fractures and fillings,the PSO-LSSVM identifying method was presented for the fractures and fillings by integrating the Particle Swarm Optimization( PSO) and the Least Squares Support Vector Machines( LSSVM). And moreover the method and BP neural network method were applied to identify the fractures and fillings of the Carboniferous carbonate reservoirs in the eastern margin of Pre-Caspian Basin. The comparison of the identified results of these two methods shows that the PSO-LSSVM method is better than the other. And furthermorethe result of the PSO-LSSVM method is well coincident with those of FMI electrical images and core data.
关键词(KeyWords)：
碳酸盐岩储层;裂缝识别;缝洞充填物;粒子群算法;最小二乘支持向量机;滨里海盆地
carbonate reservoir;fracture identification;fracture-vug filling;Particle Swarm Optimization(PSO);Least Squares Support Vector Machine(LSSVM);Pre-Caspian Basin
基金项目(Foundation): 国家高技术研究发展计划项目(2013AA064201);; 国家科技重大专项课题(2016ZX05003-003)联合资助
作者(Author):
谢玮,刘斌,钱艳苓,孙炜,史飞洲,李玉
XIE Wei,LIU Bin,QIAN Yanling,SUN Wei,SHI Feizhou,LI Yu
参考文献(References)：
- [1]卢曦.塔里木盆地古城地区下古生界碳酸盐岩成岩作用及储层孔隙特征[J].大庆石油地质与开发,2016,35(4):15-21.
- [2]谭廷栋.裂缝性油气藏测井解释模型与评价方法[M].北京:石油工业出版社,1987:164-196.
- [3]孙炜,李玉凤,付建伟,等.测井及地震裂缝识别研究进展[J].地球物理学进展,2014,29(3):1231-1242.
- [4]王晓,周文,王洋,等.新场深层致密碎屑岩储层裂缝常规测井识别[J].石油物探,2011,50(6):634-638.
- [5]徐敬领,王贵文,王亚静,等.碳酸盐岩储层测井评价方法综合研究[J].石油物探,2012,51(5):521-530.
- [6]柳建华,蔺学旻,张卫锋,等.塔河油田碳酸盐岩储层有效性测井评价实践与思考[J].石油与天然气地质,2014,35(6):950-958.
- [7]钟淑敏,綦敦科,王秀娟.应用常规测井资料识别砂泥岩储层裂缝方法[J].大庆石油地质与开发,2005,24(1):98-99.
- [8]张吉昌,邢玉忠,郑丽辉.利用人工智能技术进行裂缝识别研究[J].测井技术,2005,29(1):52-54.
- [9]贾凌霄,贾家磊,张毅,等.改进BP神经网络在裂缝预测中的应用[J].工程地球物理学报,2012,9(3):301-305.
- [10]石广仁.支持向量机在裂缝预测及含气性评价应用中的优越性[J].石油勘探与开发,2008,35(5):588-594.
- [11]瞿子易,周文,罗鑫,等.基于粒子群和支持向量机的裂缝识别[J].石油与天然气地质,2009,30(6):786-792.
- [12]田飞,金强,李阳,等.塔河油田奥陶系缝洞型储层小型缝洞及其充填物测井识别[J].石油与天然气地质,2012,33(6):900-908.
- [13]陈钢花,胡琮,曾亚丽,等.基于BP神经网络的碳酸盐岩储层缝洞充填物测井识别方法[J].石油物探,2015,54(1):99-104.
- [14]王晓兰,张万宏,王慧中.基于小波变换和AR-LSSVM的非平稳时间序列预测[J].控制与决策,2008,23(3):357-360.
- [15]Suykens J A K,Vandewalle J.Least squares support vector machine classifiers[J].Neural Processing Letters.1999,9(3):293-300.
- [16]苏建龙,米鸿,王彦春,等.基于支持向量机的非线性弹性波阻抗反演方法[J].石油地球物理勘探,2014,49(4):751-758.
- [17]Eberhart R,Kennedy J.A new optimizer using particle swarm theory[C]//Sixth Inter national Symposium on Micro Machine and Human Science.New York:IEEE,1995:39-43.
- [18]赵中平,牟小清,陈丽.滨里海盆地东缘石炭系碳酸盐岩储层主要成岩作用及控制因素分析[J].现代地质,2009,23(5):828-834.