基于PCA-BP神经网络的磨粒自动识别  

Wear Particles Identification Based on PCA and BP Neural Network

作　　者：罗炳海[1] 黄永武[2] 刘远飞[3] 

机构地区：[1]中国人民解放军驻二一二厂军事代表室 [2]中国人民解放军95389部队63分队 [3]空军工程大学工程学院

出　　处：《润滑与密封》2010年第6期117-120,共4页Lubrication Engineering

摘　　要：以航空发动机润滑油液中的磨损微粒为研究对象,提出一种采用PCA-BP分类器对磨粒类型进行自动识别的方法。该方法通过对磨粒图像进行预处理并提取一些特征参数,采用PCA对所提取的特征参数进行优选,采用BP网络分类器对磨粒类型进行计算机自动识别。结果表明,相对于传统型BP分类器的识别准确率,此分类器的识别准确率更高。The wear particles in the lubricating oil of aero-engine were researched,a wear particle recognition approach based on PCA and BP neural network was presented.This approach extracts some characteristic parameters through pretreating the wear particle images,selects the extracted characteristic parameters using PCA,realize the automatic classification of wear particles using BP neural network classifier.The classification results show that classification accuracy rate of wear particle based on PCA and BP neural network is higher than that of traditional BP neural network.

关 键 词：主成分分析法 BP神经网络 磨粒识别 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]