一种自构神经网络及其在测井资料解释中的应用

鉴于ＢＰ网络存在着学习过程收敛速度慢，网络容错能力差的缺点，本文提出了一种自构神经网络算法。该算法分两部分：１．将模糊集理论与神经网络相结合提出一个模糊动态改变学习率的有效算法；２．利用相关自动原理来动态调整网络的隐节点数，最终让其达到一个最佳的稳定状态。将此方法应用于江苏油田“镇田井”的测井资料解释，同已

Self-configuring neural network and its application in logging data interpretation

Self-configuring neural network is used because BP neural network has both slow convergence in learning course and poor fault tolerant ability. The algorithm of the network consists of two parts: Combine fuzzy set theory with neural network to form an effective algorithm in which learning rate is changed fuzzy and dynamically. Use autocorrelation principle to dynamically regulate the hidden nodes of network, which will be in optimal situation finallyThe neural network has been applied to the interpretation of logging data which were collected from Zhen-4 borehole in Jiangsu Oil Field, and brings better effect than BP neural network. The algorithm is an effective one.