Determination Of Volatile Gas In Rice During Storage By Colorimetric Sensor Array Combined Spectral Analysis Technology And Equipment Development

During transportation,sales,and distribution,the quality of rice decrease owing to environmental factors,culminating in diminished nutritional quality and safety of food.At present,the conventional detection methods are cumbersome and the detection accuracy is not high.Therefore,it is necessary to establish a set of rapid and accurate evaluation methods and detection device for estimating rice storage quality.In this study,based on the changes of volatile gases in rice during different storage periods,a colorimetric sensor array(CSA)with high sensitivity and specificity was constructed.Hyperspectral technology was used to extract the color information,followed by,the classification modeling to select a suitable information expression method.Based on this,a set of rice storage time detecting system was developed.The main research contents of this thesis are:1.Screening of characteristic gases during rice aging and optimization of reaction conditions of CSA.SPME-GC-MS was used to characterize rice volatiles at different storage times.A total of 83 odorants were detected and comparatively analyzed for rice volatile gases with extended storage time.After comparing the changes of organic volatile content,it was found that the content of n-hexanal showed a good correlation with the storage time.Therefore,n-hexanal was used as a characteristic gas for characterizing the degree of rice aging.On this basis,three kinds of fluoroboron-type color sensitive materials were selected to construct CSA,which used a characteristic gas as a main detection object and can interactively detect other rice volatile gases.The two reaction modes of CSA were optimized,and it was finally determined that the reaction was carried out 55? for 8 min using the free reaction mode.2.Detection and analysis of rice with different storage time based on colorimetric sensing-hyperspectral analysis technology.After CSA reacted with the volatile gas of rice with different storage periods,the information change of CSA was characterized by the hyperspectral technique from the image data and the spectral data.The storage time of the rice used in the experiment was 0,1,2,4,6,and 10 months and the spectral and imaging data of CSA was obtained.First,the Synergy-interval Partial Least Squares Algorithm(Si-PLS)was used to select the optimal interval of the spectral data.Then,GA,ACO and CARS was used to respectively filter spectral data.KNN and LDA were used to perform pattern recognition on the selected spectral data,it is determined that the spectral data filtered by GA-Si-PLS had the highest recognition rate and then the images at five characteristic wavelengths were selected to extract image data.Finally,a comparsion of the performance of the prediction model constructed by the spectral data,the image data and the fusion dataset of the two was performed.The results show that after the spectral data variables were extracted by GA-Si-PLS,the best prediction model was established by Linear Discriminant Analysis(LDA),when the number of principal components was 10.The correct recognition rate of the training set was found to be 92.73%,whereas rate of the correction set was 90.91%.Therefore,the spectral information is best expressed using the color information of CSA.3.Development of Rice Storage Time Detection System Based on Colorimetric SensingNear Infrared Spectroscopy Based on the results of the previous chapter,a portable intelligent system for colorimetric sensing-near infrared spectroscopy was developed to quickly and nondestructively detect rice storage time.The hardware design of the detection system was divided into a spectrum acquisition and processing module,a reaction flow control module,a reaction progress display module,a power supply module and a minimum circuit of the single chip microcomputer according to functions.The software design of the detection system includes embedded development of single chip microcomputer,serial touch screen interface writing and Windows desktop application.The assemble device used to study the different storage(0,1,2,4,and 6 months)times.Subsequently,the spectral data of CSA obtained after the reaction is extracted was then classified and modeled.The experimental results show that the spectral variables extracted by UVE-Si-PLS have the highest recognition rate after modeling with LDA algorithm.The correct recognition rate of the training set was 98.00% with 9 PCs,and the correct recognition rate of the correction set was 96.00%.The results show that the proposed CSA binding spectroscopy technique is highly suitable for the identification of rice storage time.