混沌退火Hopfield网络及其在储层聚类分析中的应用  

ANNEALED CHAOTIC HOPFIELD NETWORK AND ITS APPLICATION IN CLUSTERING OF RESERVOIR

作　　者：张军华[1] 章多荣[1] 王永刚[1] 陆文志[1] 赵勇 

机构地区：[1]中国石油大学地球资源与信息学院,东营257061 [2]Tulsa大学石油工程系,美国俄克拉荷马州74104

出　　处：《物探化探计算技术》2006年第2期89-92,共4页Computing Techniques For Geophysical and Geochemical Exploration

摘　　要：将一种基于混沌退火的Hopfield神经网络(ACHN)应用到储层聚类分析中。这里主要阐述了ACHN聚类的基本原理、混沌神经元的标定、ACHN聚类的策略、先验信息的约束以及混沌退火搜索。实际砂砾岩储层聚类分析表明:这种网络能克服普通Hopfield网络陷入局部最优的不足,得到更准确、精细的聚类结果。In this paper,an annealed chaotic Hopfield network(ACHN) is proposed and applied in clustering of the gravel body reservoir.In the paper,the basic theory and strategy of ACHN clustering are discussed in detail and how to mark chaotic neurons,constrain Hopfield network with known information,and anneal with chaos is also introduced.The much fine clustering results show that the method can successfully solve the problem of local minima in Hopfield network,and obtain the global optimal solution.

关 键 词：HOPFIELD网络 混沌神经元 退火 砂砾岩 聚类分析 

分 类 号：TP389.1[自动化与计算机技术—计算机系统结构]