Design And Research Of Electronic Nose System For Identification Of Liquor Varieties

There are many kinds of liquors in our country,and their flavor types are different.With the rapid development of economy,the development trend of the liquor industry is more and more magnificent.The profit of the liquor industry is very rich.In order to gain huge profits,some illegal traders began to produce and sell a large number of fake liquors.Drinking fake liquor is harmful to the body,so the supervision of liquor quality is particularly important.At present,the quality and variety testing of liquors mainly relies on the sensory analysis of the taster,chromatograph and mass spectrometer.Sensory analysis is subjective and susceptible to the assessor's physical condition,environment and other factors.While the chromatographs and mass spectrometers can perform accurate and objective analysis,the detection process is cumbersome,long-cycle and high-cost.In addition,the detection instruments are expensive and cumbersome,and they are not conducive to use widespreadly.Therefore,these detection methods are obviously somewhat stretched for the China's increasingly prosperous liquor industries.As a new detection method,electronic nose can identify liquors quickly according to the volatile odor of liquors.However,the existing electronic nose products are expensive and difficult to extract gas characteristic information.In addition,the electronic nose is easily interfered by environmental factors such as temperature and humidity,and its robustness is not strong.A new electronic nose system was designed with low cost,high performance,small volume and anti-jamming capability.It was developed under the support of the "Jiangsu Graduate Research and Practice Innovation Project?SJCX17₀573?".It can be used for rapid,real-time and effective identification and analysis of liquors on the market.The electronic nose system designed and developed in this paper mainly includes four parts: sensor array,data acquisition and transmission system,gas collection device and data processing.The sensor array is mainly composed of 10 kinds of gas sensors of TGS and MQ.The sensor array has the characteristics of low power consumption,high sensitivity,long life,low cost and high performance.It can obtain the odor characteristics of liquors as much as possible,and convert the characteristics into voltage signals.In the process of data acquisition,the data acquisition system?MP4623?is used.It can collect 10 channels of data at the same time,and convert the collected data by AD.Then it can also store data.At the same time,the data acquisition process can be observed in real time through LabVIEW software programming.The gas collecting device is mainly composed of two pumps,a sample storage bottle and a gas chamber.In addition,in terms of data processing,this paper combines fuzzy theory and discriminant principal component analysis?DPCA?algorithm,and proposes a fuzzy discriminant principal component analysis?FDPCA?algorithm,which can effectively improve the effect of feature extraction.FDPCA improves the accuracy of classification and recognition of electronic nose systems and achieves "soft" classification in multi-class classification.Thereby,it can improve the anti-jamming ability of electronic nose as well as the performance of electronic nose.Finally,the electronic nose system combined with the fuzzy discriminant principal component analysis?FDPCA?algorithm and K-nearest neighbor classification method,which was used to distinguish the six kinds of liquors purchased on the market.Leave one out?LOO?and kfold cross-validation method were used to verify the accuracy of classification identification.It was found that the accuracy of the identification of the six kinds of liquors could reach 98% by the electronic nose.The experimental results show that both the electronic nose system designed in this paper and the proposed FDPCA algorithm are very effective in the classification of liquors.