BP神经网络在电子鼻分类识别多品牌白酒中的应用研究

The Application Research of the BP Neural Network in Electronic Nose Classification and Recognition with Different Brands Liquor

由于MQ3、MQ4、TGS813、TGS2620 4个金属氧化物半导体组成的气体传感器阵列对酒精气体及有机物交叉敏感,据此建立实时数据采集系统,并提出稳态特征值和动态特征值的提取方法,进而结合BP神经网络识别方法,通过所建立的电子鼻系统对3种不同品牌白酒进行了分类识别实验.结果表明:电子鼻系统对不同品牌白酒的识别率稳态特征时达90.0%,动态特征时达83.3%.

Based on the sensors array made up of MQ3、MQ4、TGS813、TGS2620 four metal oxide semiconductor sensors which are cross sensitive to alcohol gas and organic,the real-time data acquisition was established in this pa-per,and the method of the steady and dynamic characteristic value extraction was proposed. Combined with BP neu-ral network recognition,three kinds of liquor were conducted classification experiments by the electronic nose sys-tem. The results show that the recognition rate of electronic nose system is up to 90. 0% under the steady character-istic and 83. 3% under the dynamic characteristic for the different brands of liquor.