Development And Analysis Of Electronic Nose Detection System For Cereal Pests

Electronic nose technology is a new gas odor recognition technology,mainly involving pattern recognition theory,sensor technology and other aspects.In recent years,with the increasing development level of electronic nose technology,the ability of electronic nose system to identify samples qualitatively and quantitatively has been enhanced,and its application has become more and more extensive.At present,electronic nose technology is mainly used in food industry,agriculture,medical treatment,environmental engineering,national defense engineering,fire protection and other fields.As a large agricultural country,China causes tens of thousands of crop losses due to insect pests every year.The traditional pest detection process of grain is subject to subjective influence,the detection cycle is long and the cost is high.Therefore,it is necessary to find a new pest detection method to control the monitoring cost.Based on the relevant theories of virtual instruments and the principle of electronic nose technology,this paper designed and realized an electronic nose system for detecting grain insect pests,which detected and analyzed grain insect pests through sensor array acquisition and a variety of pattern recognition algorithms.The specific contents of this paper are as follows:(1)The operating principle of the electronic nose was studied,and the software and hardware systems of the electronic nose were designed.The hardware system is composed of metal oxide gas sensor array,power supply device,experimental acquisition gas chamber and data acquisition system.The software system uses the virtual instrument Lab VIEW to write the upper computer program,and develops the electronic nose data acquisition experiment platform,which can display and store the response signal collected by the sensor in real time.(2)With the MATLAB algorithm analysis and SPSS software environment,complete the pretreatment of data standardization program design,with the help of principal component analysis(PCA)to extract specific features of four kinds of value,and optimize the sensor array,with the help of a linear discriminant analysis(LDA)and principal component analysis after preprocessing of the data in the form of classification recognition comparison,Support vector machine(SVM)and BP neural network were used to establish the regression model between the sensor signal and the different proportion of insect pests in rice,and the prediction was made with the model.The experimental results showed that LDA had a significant effect on the identification of rice samples infected with different amounts of Rhizoma cervae,and the accuracy of SVM model test set in predicting the doping proportion of rice infected with Rhizoma cervae was significantly better than that of BPNN model,which was as high as 91%.(3)The detection and control system based on electronic nose technology was realized,and the rice pest experiment was carried out.The system was used to classify and distinguish rice samples with different degree of insect pests.The results show that the electronic nose system can control the sampling process of the data acquisition card,the virtual instrument experimental platform can realize the human-computer interaction on the main control panel,real-time monitoring of the sampled data,data saving and other functions,and the pattern recognition algorithm can complete the task of distinguishing rice samples with different insect pests.