井层判别的人工神经元网络方法初探

FIRST PROBING ON ARTIFICIAL NEURAL NETWORK METHOD FOR LAYER RECOGNITION IN WELL-LOGGING INTERPRETATION

测井综合解释过程中的井层判别可看成是一个模式识别和分类问题。人工神经元网络具有很好的分类、学习和容错能力。本文首先介绍人工神经元网络的研究现状,着重介绍向后传递网络(B—P网络)的基本原理。然后,本文提出了进行输入变换的方法来解决网络的收敛问题。最后本文给出利用人工神经元网络进行井层判别的一些试验结果。

The layer recognition in the process of integrated logging interpretation can be regarded as a classification problem based on pattern recognition. It's known to us that artificial neural networks work very well in operating classification and learning and it's tolerant of errors as well. This paper will first introduce the current situation of studying artificial networks with the emphasis on the fundamentals of the back-propagation network. A method of input swapping is then presented to deal with the problem of network covergence. Some results from the experiment on applying the artificial neural network method to layer recognition are given at the end of the paper.