利用人工蜂群算法进行地震属性聚类分析  

Seismic attribute clustering based on artificial bee colony algorithm

作　　者：曹成寅[1] 郑晓东[1] 李艳东[1] 李劲松[1] 

机构地区：[1]中国石油勘探开发研究院,北京100083

出　　处：《石油地球物理勘探》2015年第4期684-690,5,共7页Oil Geophysical Prospecting

基　　金：国家重大专项课题"海相碳酸盐岩储层地震描述与油气藏有效预测技术研究(二期)"(2011ZX05004-003);中国石油勘探开发研究院科学研究与技术开发项目"数据驱动的非线性智能地震储层预测技术研究"(2012Y-013)联合资助

摘　　要：将具有全局搜索能力的人工蜂群算法引入到地震属性聚类中,在人工蜂群算法中蜜蜂被分为引领蜂、跟随蜂和侦察蜂三种,引领蜂寻找蜜源,跟随蜂根据引领蜂带回的蜜源信息搜索新蜜源,若某个蜜源被引领蜂或跟随蜂丢弃,则侦察蜂随机搜索新蜜源代替被丢弃的蜜源。其中每个蜜源代表优化问题的一个可能解,蜜源的花蜜量对应于解的质量,用适应值表示。最优解的搜索过程包括初始化和蜜源更新两个过程,通过蜂群分工协作,搜索最佳聚类中心,解决常规K均值聚类算法存在的多解性问题。利用国际上通用UCI机器学习数据库中的Iris和Wine数据集对常规K均值算法和人工蜂群算法进行了测试,结果表明:基于人工蜂群算法的聚类方法的分类正确率明显高于K均值方法。实际地震数据的聚类结果表明:基于人工蜂群算法的地震属性聚类能精细刻画塔北奥陶系碳酸盐岩岩溶储层的相带特征,并且能较好地描述呈片状弱反射特征的储层分布。The artificial bee colony(ABC)algorithm is introduced into seismic attribute clustering in this paper.In the ABC algorithm,there are three types of bees:employed bees,onlooker bees and scouter bees.The employed bees search food sources,onlooker bees search new food sources based on the information shared by employed bees.If a food source is abandoned,scouter bees search randomly a new food source to replace the abandoned one.In the ABC algorithm,every food source represents a possible solution,and the amount of nectar in the food source represented by fitness corresponds to the quality of the solution.The process of finding the optimized solution includes two steps:initialization and food source update.Through the cooperation of bee colony to search the best clustering centroids,the problem of multiple solutions could be solved in ordinary clustering algorithm.Tests on Iris and Wine datasets in the machine learning UCI dataset bank proved the clustering method based on ABC has higher accuracy than K-means.For seismic data tests,seismic attributes clustering based on ABC can accurately describe Karst reservoir microfacies of Ordovician carbonate and char-acterize weak flake-like reflection reservoirs in North Tarim Basin.

关 键 词：地震属性聚类 人工蜂群算法 串珠状强反射特征 片状弱反射特征 局部最优 最优聚类中心 

分 类 号：P631[天文地球—地质矿产勘探]