摘要
针对测井曲线小层水淹层自动识别,提出一种基于自组织过程神经网络(Self-Organization PNN,SOPNN)的动态判别模型和方法.SOPNN由输入层和竞争层组成,其输入和连接权为与时间有关的函数.网络可将小层连续测井曲线作为输入,自动提取曲线所对应的形态和幅值特征,并对其进行自组织,在竞争层将分类结果表示出来.给出SOPNN竞争学习和有教师示教2种学习算法,对实际测井水淹层资料进行处理,正确识别率可达79.25%.
Aiming at the problem for automatic recognition of logging curves in thin water-flooded layer,a dynamic classification model and method based on self-organization process neural networks(SOPNN)is presented.The SOPNN consists of input layer and competitive layer,whose inputs and connection weights can be functions of time.The continuous logging curves in thin layers can be inputs of the network,which extracts the corresponding forms and amplitude values automatically,implements self-organization and then outputs the classification result in the competitive layer.The learning algorithms of SOPNN competitive learning and teacher demonstration are given in this paper,which are used to process actual data of water-flooded layer logging and obtain good results.
出处
《大庆石油学院学报》
CAS
北大核心
2008年第6期97-99,共3页
Journal of Daqing Petroleum Institute
基金
黑龙江省自然科学基金项目(ZA2006-11)
黑龙江省科技攻关项目(GZ07A103)
黑龙江省普通高等学校骨干教师创新能力计划项目(105G002)
关键词
SOPNN
学习算法
模式识别
测井曲线
水淹层
SOPNN
learning algorithm
pattern recognition
logging curve
water-flooded layer
