基于电子鼻和电子舌信息融合的白酒质量检测方法研究  

Research on Liquor Quality Detection Method Based on Information Fusion of the Electronic Nose and the Electronic Tongue

作　　者：马泽亮 程琦峰[1] 程盟盟 杜岗 MA Zeliang;CHENG Qifeng;CHENG Mengmeng;DU Gang(School of Information Engineering,Lianyungang Technical College,Lianyungang 222006,China;School of Mechanical and Electrical Engineering,Lianyungang Technical College,Lianyungang 222006,China)

机构地区：[1]连云港职业技术学院信息工程学院,江苏连云港222006 [2]连云港职业技术学院机电工程学院,江苏连云港222006

出　　处：《连云港职业技术学院学报》2022年第2期14-18,共5页Journal of Lianyungang Technical College

基　　金：连云港市科技计划项目(SF2134);连云港市应用研究重大课题青年项目(SLYZ21408)。

摘　　要：自行开发了一种可检测白酒纯度的新型的电子鼻和电子舌检测仪器,利用特征级信息融合的方式将电子鼻和电子舌信号进行数据融合,采用极限学习机(Extreme Learning Machine,ELM)进行白酒纯度辨别分析,使用人工蜂群优化最小二乘支持向量机(Artificial Bee Colony Least Squares Support Vector Machine,ABC-LSSVM)进行白酒纯度预测分析。结果表明:只使用电子鼻或电子舌对白酒纯度辨别时准确率均为93.3%,将电子鼻和电子舌的特征数据进行融合后分辨正确率得到了有效提高;在对白酒纯度预测时,将电子鼻和电子舌数据融合后利用模式识别算法处理和只使用电子鼻或电子舌数据进行处理相比,验证集均方根误差下降,相关系数上升,预测结果更接近真实值,进一步表明嗅觉和味觉密不可分,将两者进行数据融合能够大幅度提高分辨正确率和预测精度。A new detection instrument that can detect the purity of liquor has been developed based on the electronic nose and the electronic tongue. It conducted data fusion of the electronic nose and the electronic tongue signals by using feature-level information fusion,the ELM(Extreme Learning Machine)for the liquor purity discrimination analysis,and the ABC-LSSVM(Artificial Bee Colony Least Squares Support Vector Machine)for liquor purity predictive analysis. The findings show that the accuracy of liquor purity is 93.3% by only using the electronic nose and the electronic tongue and the resolution accuracy of electronic nose and electronic tongue is effectively improved after using feature-level information fusion. In the liquor purity prediction,compared to data processing by the electronic nose or the electronic, data fusion after using pattern recognition algorithm lowers the root mean square error of the validation set,increases the correlation coefficient,and makes the prediction results closer to the true value. It further shows that the smell and taste is inseparable, and that the data fusion can greatly improve the discrimination accuracy and prediction accuracy.

关 键 词：电子鼻 电子舌 信息融合 极限学习机 支持向量机 

分 类 号：TS262.3[轻工技术与工程—发酵工程]