反向传播神经网络模型及其在测井资料岩性自动识别中的应用

BACK PROPAGATION NEURAL NETWORK MODEL AND ITS APPLICATION TO AUTOMATIC LITHOLOGIC IDENTIFICATION IN WELL LOGGING

地矿部石油物探研究所,南京210014

The Research Institute of Geophysical Prosepcting for Petroleum, MGMR, Nanjing 210014

本文首先介绍反向传播(BP)神经网络模型,包括神经元的结构、反向传播神经网络的结构、反向传播神经网络的学习算法和运行。然后介绍根据测井资料,应用反向传播神经网络模型作岩性自动识别的方法,包括网络结构参数的选择、学习样本的选取、网络学习和运行。最后,我们对二个实例进行处理解释,取得了比较满意的解释结果。

This paper begins with an introduction on back propagation neural network models including the structures of neural cells and the back propagation neural network as well as its learning algorithms and execution,then proceeds to study the application of the model to well logging data for lithologic identification including the selection of network structure paramaters and training samples as well as the learning and operation of networks. At last, two data processing examples are cited to show the satisfactory results achieved by applying the method.