地质样品的分类是地质学中一个重要问题。本文提出用一种神经网络———ART2模型来实现自动分类。该方法简单、实用 ,分类效果令人满意。网络的应用结果显示出它具有良好的获得最佳分类数和最佳分类的双重功能 The classification of geological samples is an important issue in geology. This paper proposes to use a neural network --- ART2 model to achieve automatic classification. The method is simple and practical, the classification effect is satisfactory. The results of the network show that it has a good dual function of getting the best number of categories and best classification