摘要
介绍了神经网络的基本成分-神经元对信号的接收、激活与处理输出过程和网络的结构。从目前已有的几种模型中选取了误差反向传播网络模型。阐明了使用该模型进行训练与预测的方法和步骤,并用于测井中的粒度中值和渗透率解释。其预测结果、平均相对误差分别是7.84％与22.98％,而用以前其它方法,这一误差分别是20％与40％左右。
The basic features of nerve network and its implementation process are presented in this paper. The selected BP network model is shown in detail and applied to interpretation of the median grain diameter and permeability in logging. The mean relative errors of the predicted results are 7.84% and 22.98% respectively, but we use previous method, these errors are about 20% and 40% respectively.
出处
《河南石油》
1996年第2期21-25,5,共5页
Henan Petroleum
关键词
测井解释
神经网络
粒度
渗透率
数据处理
Well logging, Interpretation, Nerve network, Data processing
