摘要
根据芯片加工完成后常见缺陷,提取特征并将其分类。基于信息熵特征选择方法,将这些特征量进行优化选择。基于BP神经网络的构建了筛选器模型,并设计了机械结构,通过现场对样本的采集进行训练与测试,结果表明该筛选器能够很好识别5种常见的芯片缺陷。
According to common-defects of the chips that had been produced , distilling features and classifying them.In view of the method that was used to choose information entropy features ,this features were optimized for choosing.In view of BP neural network, constructing the model of selectors, and designing the mechanical structure, training and testing the results of taking samples, indicating that the selector can recognize 5 types of chip common-defect well.
出处
《装备制造技术》
2015年第2期63-65,共3页
Equipment Manufacturing Technology
关键词
筛选器
信息熵
模式识别
filter
information entropy
pattern recognition