摘要
概率神经网络(PNN)实际上是一种数学插值手段,它的实现引用了神经网络的结构。但它通过研究数学规则,可以比多层前馈神经网络更好地实现插值功能,因而具有潜在的优势。这里简要介绍了PNN的基本原理并用于储层物性参数的预测。实践表明该方法是有效的。
PNN is a mathematical interpolation method in fact. It's realization uses the figure of neural network, but it can realize the interpolation more effective than multilayer forward feedback neural networks, therefore it has potential advantage. The paper introduces simply the basic principle of PNN and its application in prediction of reservoir parameters. It turns out that the method is effective.
出处
《物探化探计算技术》
CAS
CSCD
2003年第1期78-82,共5页
Computing Techniques For Geophysical and Geochemical Exploration
