Kohonen网络在奥灰岩溶发育带横向预测中的应用
|
摘要
自组织Ｋｏｈｏｎｅｎ网络是一种学习速度很快的神经网络，可以用于分类、聚类、解释等问题。本文依据奥灰岩地震波运动学和动力学特征，提取时间域最大互相关系数、分形关联维、频率域主频、频带宽度和主频带能量共５个参数，利用自组织（Ｓｅｌｆ－Ｏｒｇａｎｉｚｉｎｇ）Ｋｏｈｏｎｅｎ人工神经网络横向预测含水裂隙发育带。试算结果表明，方法可行，可望成为预测奥灰岩岩溶裂隙发育带的一种有效方法。
Self-organization Kohonen network is a fast learning neural network used to deal with problems of classification,clustering,interpretation and so on.This paper derived five parameters such as maximum crosscorrelation coefficient,fractal associative dimension in time domain,and dominant frequency,bandwidth and dominant energy in frequency domain according to the seismic kinematics and dynamic characteristics of Ordovician limestone.It made use of the self-organization Kohonen artificial neural network to predicate laterally the aqueous fractured zone.Experiments on real seismic data have showed that the technique was feasible.It cad become an effective method to predicate the karst fractured zone in Ordovician limestone.
Self-organization Kohonen network is a fast learning neural network used to deal with problems of classification,clustering,interpretation and so on.This paper derived five parameters such as maximum crosscorrelation coefficient,fractal associative dimension in time domain,and dominant frequency,bandwidth and dominant energy in frequency domain according to the seismic kinematics and dynamic characteristics of Ordovician limestone.It made use of the self-organization Kohonen artificial neural network to predicate laterally the aqueous fractured zone.Experiments on real seismic data have showed that the technique was feasible.It cad become an effective method to predicate the karst fractured zone in Ordovician limestone.
引文
1张际先，宓霞．神经网络及其在工程中的应用．北京：机械工业出版社，1996；84～912蔡煜东，杨兵，汤军彪．自组织人工神经网络在煤层对比判别中的应用．物探化探计算技术，1995；17（4）：81～873刘铃，黄铃，张晓东等．用自组织映射方法进行油气检测．石油物探，1994，33（4）：56～644陈遵德，朱广生．Kohonen网络在油气横向预测中的应用．石油物探，1995；34（2）：53～565蔡煜东，杨兵，汤军彪．自组织人工神经网络预测碳酸岩中岩溶洞穴规模．物探化探计算技术，1995；17（2）：40～44
版权所有：© 2023 中国地质图书馆 中国地质调查局地学文献中心