一类正则模糊神经网络及在沉积微相识别中的应用  

A normal fuzzy neural network and its application to sedimentary facies identification

作　　者：许少华[1] 何新贵[2] 梁久祯[1] 

机构地区：[1]大庆石油学院计算机系,黑龙江安达151400 [2]北京系统工程研究所,北京100101

出　　处：《控制与决策》2002年第3期332-335,共4页Control and Decision

基　　金：黑龙江省教育厅科学技术基金项目 (10 5 11119)

摘　　要：考虑一种 5层结构的正则化模糊神经网络模型 ,针对网络结构的优化问题给出了该网络模型规则层节点的选取方法和相应的反传学习规则 ;针对样本筛选问题 ,提出一种按模糊隶属函数值相近样本向量类别矫正策略。将正则模糊神经网络用于油藏沉积微相的识别 ,可自动提取测井曲线与微相类型之间复杂的映射关系 ,实现沉积微相的连续识别。实际资料处理结果表明 。A normal fuzzy neural network(NFNN) with five layers is proposed. Focusing on the structure optimization of network, a new node selection method and corresponding back propagation learning algorithm are presented. Considering contradiction and superfluous in a pattern set that may be brought by rough selection, a strategy in which some pattern′s regimentation is remedied according to their fuzzy value is proposed. NFNN is applied to sedimentary facies identification to find the relation between logging curve and sedimentary facies classifications. Test results illustrate its practicability.

关 键 词：模糊神经网络 学习算法 模式识别 正则化 微相识别 油藏沉积 油田勘探 

分 类 号：TE19[石油与天然气工程—油气勘探] TP183[自动化与计算机技术—控制理论与控制工程]