模糊竞争神经网络在地震微相划分中的应用

为了满足地震解释中无监督型聚类分析的需要,我们提出了一种模糊自组织神经网络的结构和算法,它是在Kohonen自组织网、模糊网及竞争网的基础上发展起来的,它将模糊自组织神经网络中的隶属度概念引入了网络的训练过程,因而收敛速度很快;同时借鉴了竞争网良好的自适应和动态聚类能力及Kohonen网的映射密度图的概念,因此它具有快速处理大容量的数据且分类精度易于控制的特点。此种算法用于地震微相的划分取得良好的效果。

APPLICATION OF THE COMPETITIVE FUZZY NEURAL NETWORK TO SEISMIC MICROFACES ANALYSIS

To meet the demand of clustering analysis without supervision in seismic interpretation, a new type of neural network which is called competitive fuzzy neural network (CFNN) developed.This kind of neural network, which was developed on the basis of Kohonen, fuzzy and competitive neural network, the grade of membership was introduced to its course of network training, and thus can converge rapidly.In addition, the CFNN absorbed the competitive network the advantage of self-organizing and dynamically clustering, it characterized by fast processing a large amount of data and the abillty to flexibly control the numbers of final clustering.As applying it to seismic microfaces analysis,a good result is obtained.