基于单亲遗传算法的模式聚类在复杂岩性识别中的应用  

Application of a partheno-genetic algorithm pattern clustering to complicated lithology identification

作　　者：许少华[1] 孙文德[2] 肖玉灿 袁新强[3] 王焕琴 

机构地区：[1]大庆石油学院计算机科学系,黑龙江安达151400 [2]大庆油田有限责任公司第三采油厂,黑龙江大庆163113 [3]大庆油田有限责任公司第二采油厂,黑龙江大庆163414

出　　处：《大庆石油学院学报》2000年第3期54-56,共3页Journal of Daqing Petroleum Institute

基　　金：黑龙江省自然科学基金!资助项目 (F9931)

摘　　要：针对油藏地质中复杂地层岩性划分存在的模糊性和多解性 ,提出了一种基于单亲遗传算法的模式聚类划分复杂地层岩性的方法 .该方法通过隐含传统算法交叉算子功能的基因换位等遗传算子来实现进化操作 ,简化了遗传过程 ,同时避免了早熟收敛问题 .用PGA求解模式聚类问题可使聚类结果不依赖于初始聚类中心 .采用该方法对大庆油田10 8个地层样本进行处理 ,总符合率达 88.0 %.结果表明该方法可提高复杂岩性识别的精度 .Considering the classification indetermination and diversity in petroleum reservoir geology's complicated lithology partition, a method based on Partheno\|Genetic Algorithm's pattern clustering is proposed in this paper. This method realizes evolving operation by genetic operators such as genetic transposition, genetic shift and so on, which hide traditional Genetic Algorithm crossover operators' function, predigest heredity process and avoid prematurity convergence at the same time. Using PGA to solve pattern clustering problem make clustering's result independent of initial clustering. By coping with Daqing oil field's real data, the method is proved effective.

关 键 词：亲单遗传算法 模式聚类 岩性划分 地层 油藏地质 

分 类 号：TE121.34[石油与天然气工程—油气勘探] P618.130.2[天文地球—矿床学]