西南石油大学学报(自然科学版)
西南石油大學學報(自然科學版)
서남석유대학학보(자연과학판)
JOURNAL OF SOUTHWEST PETROLEUM UNIVERSITY(SEIENCE & TECHNOLOGY EDITION)
2014年
2期
71-78
,共8页
庞河清%匡建超%蔡左花%廖开贵%王众
龐河清%劻建超%蔡左花%廖開貴%王衆
방하청%광건초%채좌화%료개귀%왕음
粒子群算法%核主成分分析%支持向量机%气水层识别%新场须二气藏
粒子群算法%覈主成分分析%支持嚮量機%氣水層識彆%新場鬚二氣藏
입자군산법%핵주성분분석%지지향량궤%기수층식별%신장수이기장
particle swarm optimization%kernel principal component analysis%support vector machine%gas-water layer iden-tification%Xu 2 Member gas reservoir of Xinchang Gas Field
川西新场气田须二气藏为典型的低渗致密碎屑岩气藏,由于地质条件复杂,储层非均质性严重,气水分布十分复杂,束缚水含量较高,气层、气水同层电阻率界限模糊不清,测井解释往往造成很大误判。针对这一难点,应用基于粒子群算法(PSO)的核主成分分析与支持向量机(KPCA-SVM)模型进行气水层识别。模型先通过核主成分分析(KPCA)进行非线性属性变量提取,再将提取的属性变量作为支持向量机(SVM)的输入变量,在识别过程中利用粒子群算法(PSO)寻优,最终实现气水层识别。将模型应用于新场气田须二气藏气水层识别,识别结果符合研究区的实际情况。
川西新場氣田鬚二氣藏為典型的低滲緻密碎屑巖氣藏,由于地質條件複雜,儲層非均質性嚴重,氣水分佈十分複雜,束縳水含量較高,氣層、氣水同層電阻率界限模糊不清,測井解釋往往造成很大誤判。針對這一難點,應用基于粒子群算法(PSO)的覈主成分分析與支持嚮量機(KPCA-SVM)模型進行氣水層識彆。模型先通過覈主成分分析(KPCA)進行非線性屬性變量提取,再將提取的屬性變量作為支持嚮量機(SVM)的輸入變量,在識彆過程中利用粒子群算法(PSO)尋優,最終實現氣水層識彆。將模型應用于新場氣田鬚二氣藏氣水層識彆,識彆結果符閤研究區的實際情況。
천서신장기전수이기장위전형적저삼치밀쇄설암기장,유우지질조건복잡,저층비균질성엄중,기수분포십분복잡,속박수함량교고,기층、기수동층전조솔계한모호불청,측정해석왕왕조성흔대오판。침대저일난점,응용기우입자군산법(PSO)적핵주성분분석여지지향량궤(KPCA-SVM)모형진행기수층식별。모형선통과핵주성분분석(KPCA)진행비선성속성변량제취,재장제취적속성변량작위지지향량궤(SVM)적수입변량,재식별과정중이용입자군산법(PSO)심우,최종실현기수층식별。장모형응용우신장기전수이기장기수층식별,식별결과부합연구구적실제정황。
Xu 2 Gas Reservoir,which is in Xinchang Gas Field in western Sichuan Basin,is a typical low-permeability and tight clastic gas reservoir. Due to the complicated geological conditions and serious heterogeneity in this area,the gas-water layer distribution is very complicated,and the bound water’s content is high. The boundaries of resistivity between gas reservoir and gas-water layer are blurred,so that some mistakes arise in log interpretation. We use kernel principal component analysis and support vector machine,also known as KPCA-SVM model,which is based on particle swarm optimization(PSO),to solve the problem. Firstly,the model extracts non-linear properties of variables by kernel principal component analysis(KPCA), and then inputs the properties of a variable into the support vector machine(SVM). And in the identification process,we use the particle swarm optimization(PSO)to seek the optimization algorithm. Finally,the gas-water layer identification is implemented in the SVM. We applied this model to gas&water layer prediction of Xu 2 Member gas reservoir of Xinchang Gas Field,and the recognition result is in line with the actual situation of the study area.