用于苹果品质监测的多温度扫描电子鼻系统  

Multi-temperature scanning electronic nose system for apple quality monitoring

作　　者：刘旭 余隽 李中洲 魏广芬[2] 黄正兴 朱慧超 LIU Xu;YU Jun;LI Zhongzhou;WEI Guangfen;HUANG Zhengxing;ZHU Huichao(Liaoning Key Laboratory of Integrated Circuit and Biomedical Electronic System,Faculty of Electronic Information&Electrical Engineering,Dalian University of Technology,Dalian 116024,China;School of Information and Electronic Engineering,Shandong Technology and Business University,Yantai 264005,China)

机构地区：[1]大连理工大学,电子信息与电气工程学部,辽宁省集成电路与生物医学电子系统重点实验室,辽宁大连116024 [2]山东工商学院信息与电子工程学院,山东烟台264005

出　　处：《传感器与微系统》2023年第10期72-76,共5页Transducer and Microsystem Technologies

基　　金：国家自然科学基金资助项目(61874018);山东省自然科学基金资助项目(ZR2019MF069)。

摘　　要：基于互补金属氧化物半导体(CMOS)集成微气体传感器阵列芯片开发了一种用于苹果品质监测的便携式电子鼻系统。使用2 mm×2 mm芯片上集成的4只可独立控温的半导体气体传感器,对品质良好、有瑕疵以及两者混合的苹果样本进行了气体检测。通过在传感器气敏性能良好的150~350℃温区内以50℃为步长进行温度扫描,构成虚拟传感器阵列。对其在不同温度稳态条件下提取的特征混合样本,进行了显著性分析,从而筛选出了不同气氛显著性差异大的4个温度特征样本。对该样本采用线性判别分析(LDA)算法降维,然后采用K最近邻(KNN)算法对苹果样品进行分类。结果表明:所提出的多温度稳态特征混合方法的最高分类准确率为98.3%,与单一和非稳态温度结果相比得到了大幅提升。A portable electronic nose system based on the complementary metal-oxide semiconductor(CMOS)chip integrates micro gas sensor array is developed for apple quality monitoring.Apple samples of good quality,flaws and a mixture of the two are measured using 4 independently temperature-controllable semiconductor gas sensors integrated on the 2 mm×2 mm chip.A virtual sensor array is formed by performing temperature scanning in a temperature range of 150℃to 350℃with a step length of 50℃under the condition of the gas sensitive response well.It is found that the eigenvalue samples at different steady-state temperatures with significant differences in different atmospheres by performing significance analysis on the mixed samples of the virtual sensor array.Then,linear discriminant analysis(LDA)and K-nearest neighbor(KNN)algorithm classification are performed on the samples for dimensionality reduction and classification.The results show that,compared with the single and non-steady-state temperature results,the highest classification accuracy of the proposed multi-temperature steady-state feature hybrid method reaches 98.3%,which is improved greatly.

关 键 词：电子鼻 苹果品质监测 多温度扫描 线性判别分析 K最近邻 

分 类 号：TP212.9[自动化与计算机技术—检测技术与自动化装置]