摘要
针对具有一定相关性气味信号的识别分类问题,设计了一种识别分类模型。使用PEN3电子鼻传感器阵列采集气味信号,判断所采集的气味特征参数间是否具有一定相关性。在对具有一定相关性的气味特征参数进行主成分分析降维的基础上,运用欧式距离分类法进行气味分类识别。最后通过不同品质的牛奶气味验证了算法的有效性,并与电子鼻软件的欧式距离算法进行了对比。结果显示,基于主成分分析的欧式距离算法对牛奶气味的识别效果更好。
In view of the recognition and classification of the odor signals with certain correlation, a classification model is designed. Using sensor array of the PEN3 electronic nose to collect odor signals, and determine whether there is a certain correlation between the odor characteristic parameters. Based on the principal component analysis of odor characteristic parameters, the classification and recognition of odor by Euclidean distance classification method was used. Finally, the effectiveness of the algorithm is verified by the odor of milk which are different quality, and the algorithm is compared with the Euclidean distance which is in the electronic nose＇s software. The results show that the Euclidean distance algorithm based on principal component analysis is better for the identification of milk smell.
出处
《电子设计工程》
2017年第13期1-4,共4页
Electronic Design Engineering
基金
内蒙古自治区自然科学基金(2014MS0619)
内蒙古自治区科技计划项目(20120304)
关键词
气味信号
电子鼻
主成分降维
欧式距离
odor signal
electronic nose
principal component dimension reduction
euclidean distance
