摘要
本文介绍了 BP 神经网络的基本原理以及磨粒显微图像的计算机处理与分析方法。将图像处理与神经网络技术应用于铁谱分析中的4类典型磨粒识别,提取磨粒的特征参数,训练磨粒识别神经网络,可实现4类典型磨粒的计算机识别,其识别率令人满意。
In this paper,the basic principle of the BP neural network and the com- puter image processing technique of the microscopic image of wear particles are described. The image processing and neural network technology are applied to ferrographic analysis for wear particle identification,extracting the features on particles and training the neural net- work for wear particle identification,which can automatically identify four typical worn-off particles.The experimental results are satisfactory.
基金
中国汽车工业公司"八.五"攻关项目
