Research On Electronic Nose System Based On FBAR Sensor Array

Volatile organic compounds,commonly expressed as VOCs,are important precursors for the formation of secondary pollutants such as fine particles(PM2.5)and ozone.Most VOCs have special odor,especially benzene,toluene,etc.,which will cause great harm to human health,and these odor will cause damage to human respiratory system.Therefore,it is of great significance to study the detection and recognition system of toxic and harmful gases for environmental monitoring,disease diagnosis and other fields.The system can accurately predict the types and concentrations of gases,so as to ensure that people live in a healthy environment.VOCs are toxic,irritant,teratogenic and carcinogenic,Although a single gas sensor can detect a single gas timely and reliably,it usually has the problem of poor single and repeatability.Therefore,the performance of a single device often can not meet the actual needs.The electronic nose based on sensor array can identify single or mixed gas and odor in real time and accurately,which is considered as an important development direction of VOCs gas monitoring.Gas data analysis based on FBAR array:using different polymer coated FBAR to form sensor array to test the gas,and then analyze the sensitivity,selectivity,humidity and other performance of the collected gas.Based on the principle of mass sensitivity,this paper designs an electronic nose scheme of FBAR sensor array combined with pattern recognition technology,which is used for qualitative recognition and quantitative detection of VOCs.The paper mainly includes the following points:(1)Construction of electronic nose test system and analysis of gas data:the system is composed of gas cleaning,drying pipe,valve,test room,FBAR sensor array,pattern recognition algorithm software made by PC and professional data analysis software.The FBAR sensor array consists of four kinds of polymer coated FBAR,which can avoid the interference of other gases to the target gas and accurately identify the target gas in the complex working environment.Then the system is used to sample and analyze the gas.(2)Based on image analysis and principal component analysis(PCA),the gas is identified legally:image analysis is to form specific iconic figures with the parameters of gas frequency shift and concentration,and recognize the gas with the help of the visual differences of these figures.The identified gas presents different columns on the sensor array,forming a specific gas histogram QR code,which can be used as fingerprint for VOCs classification parameter identification and VOCs classification.Principal component analysis will use maltab software to analyze the data,transform and reduce the dimension of the data,and classify the feature vectors linearly after dimension reduction.Finally,the main two-dimensional scatter diagram will be displayed on the scatter diagram of PCA analysis.The cumulative variance contribution rate of the first two principal components(PC)is 99%,so different gases can be distinguished well.(3)Qualitative identification and quantitative detection of gas by BP neural network and PCA+BP neural network:BP neural network is used to predict the type and concentration of single gas and mixed gas.In the analysis and establishment of BP neural network in MATLAB,the first thing we need to do is to normalize the data to prevent the measurement error caused by the quantity level.Then we explore the influence of the number of hidden layer neurons,activation function,performance target and other structural parameters of BP neural network on the network prediction performance,and debug the best structural parameters for this paper.Finally,BP neural network and PCA+BP neural network are used to identify and detect gas qualitatively.The research shows that::the electronic nose system has the characteristics of high sensitivity,good selectivity,fast response time/recovery time,etc.the recognition method used can effectively predict the gas:Based on the simple and visual classification of gas by principal component analysis and graphic recognition,the overall recognition rate of single gas by BP neural network is 91.7%,and the average concentration error of four single gases is detected The results show that the electronic nose system is stable and reliable in operation,fast in detection speed and high in recognition accuracy.It can meet the design expectation well and has a wide application prospect in gas environment detection.