灰铸铁石墨形态的自动分类

Auto-classification to Graphite Morphology of Gray Cast Iron

在所提取的纹理特征基础上，使用误差后向传播神经网络构建了一种优化的人工神经网络分类器，实现了灰铸铁石墨形态的自动分类。用于描述石墨形态的特征由分形维、粗细参数和二维自回归系数共同组成。该法成功地将人工神经网络引入了对灰铸铁石墨形态的分类，相对于传统人工目测法是一种很大的进步，而神经网络分类器的优化方法对其它神经网络模型的构建也具有一定参考价值。

A back propagation artificial neural network(BPANN) classifier of gray cast iron graphite mor- phology is developed, based on its texture feature, and auto-classification of gray cast iron graphite morphology is realized. Parameters describing character of gray cast graphite morphology such as fractal dimension, roughness and regression coefficients were used to the classification. The method introduces ANN into the analysis of gray cast ha graphite morphology successfully, and move greatly ahead com- paring with traditional visual observation. The optimization procedure of ANN classifier is also referential to building of other neural network models.