| 神经网络用于岩性及岩相预测的可行性分析 |
| |
| 引用本文： | 邱颖,孟庆武,李悌,张立志.神经网络用于岩性及岩相预测的可行性分析[J].地球物理学进展,2001,16(3):76-84. |
| |
| 作者姓名： | 邱颖 孟庆武 李悌 张立志 |
| |
| 作者单位： | 1. 大庆石油管理局勘探开发研究院,
2. 中国科学院软件研究所, |
| |
| 基金项目： | 国家自然科学基金“九五”油储项目（49894190）资助． |
| |
| 摘 要： | 数学可解性是利用神经网络算法求解问题需要首先考虑的问题，其次是用于训练的数据集有效性。本文针对地层沉积相预测问题，从网络映射分析角度计算了三层神经网络容量能力，求证了隐层节点与网络稳定性的关系，并给出内在关系式，从计算能力上分析神经网络用于岩性及岩相预测的可行性，为克服神经网络映射的复杂性和训练数据的不确定性提供理论依据。
|
| 关 键 词： | 网络映射方程 映射秩 网络容量 隐层节点 地层沉积相 |
| 文章编号： | 1004-2903(2001)03-0076-09 |
| 修稿时间： | 2000年1月8日 |
THE FEASIBILITY ANALYSIS OF NEURAL NETWORK IN THE PREDICTION OF LITHOLOGY AND LITHOFACIES |
| |
| QIN Ying,MENG Qing-wu,LI Ti ZHANG Li-zhi.THE FEASIBILITY ANALYSIS OF NEURAL NETWORK IN THE PREDICTION OF LITHOLOGY AND LITHOFACIES[J].Progress in Geophysics,2001,16(3):76-84. |
| |
| Authors: | QIN Ying MENG Qing-wu LI Ti ZHANG Li-zhi |
| |
| Abstract: | Mathematical solubility should be first considered when using neural network algorithm to find solutions, and the next in order is the effectivity of data set which used in training. In this paper, we compute the capacity of three level neutral network for the predication of sediment lithofacies in re- spect of a network mapping, also we proof the relation of the hidden node number, network stability and given an internal relationship which analysis the feasibility of neural network in the predication of lithology and lithofacies so as to offer a theory criterion for overcoming the complexity of neural network and the indeterminacy of training data. |
| |
| Keywords: | Network mapping equation Quadratic form Mapping rank Network capacity Predica- tion of lithology and lithofacies |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
