Quality Detection And Variety Discrimination Of Wheat Flour Based On Infrared Photoacoustic Spectroscopy

Fourier transform infrared photoacoustic spectroscopy(FTIR-PAS)is a combination of photoacoustic spectroscopy and Fourier transform infrared spectroscopy.Compared with transmissive and reflective spectroscopy,photoacoustic spectroscopy acquires the amount of spectral absorption directly,reduces the interference of scattered light and refracted light on the detector,and can greatly improve the accuracy of detection.The application ranges are wider,than suitable for solid-liquid or the analysis of powder and transparent or opaque samples.It is not affected by the particle size of the sample,does’t require pretreatment of the sample,requires less sample,can be truly non-destructive testing.The use of Fourier transform infrared photoacoustic spectroscopy(FTIR-PAS)for rapid nondestructive quality inspection and evaluation of agricultural products is one of the originally expanded application fields of infrared photoacoustic spectroscopy.Wheat is one of the most important cereals in the world.It plays an important role in agricultural and industry production in our country.In the process of production and processing,wheat grains usually enter the market or food processing after grinding and milling,and then classify their uses and later processing according to their quality characteristics.Therefore,the development of non-destructive rapid detection technology for wheat quality is of great significance in wheat production,market transactions,food processing and other links.In order to exploit a new method of wheat quality rapid non-destructive measuring technology,140 wheat flour of 14 varieties including ’Qianmai 2’,’Heijin 2’,’Luozhen’,’Heixiaomai 1’,’Zhongmai 895’,’Bainong 207’,’Zhoumai 26’,’Henong 416’,’Zhoumai 32,’Jimai 22’,’Luyuan 50’,’Shannong 20’ and ’Jiangmai 207’ were used as test samples,and the signal collected by fourier transform infrared photoacoustic spectroscopy(FTIR-PAS).The content of protein,wet gluten and ash in wheat flour was determined by national standard method as reference value,and a quality detection and variety discrimination prediction model based on FTIR-PAS was established.At the same time,in order to verify the feasibility of detection and discrimination by photoacoustic spectroscopy technology,fourier transform mid-infrared(FTIR-MIR)and near infrared(NIR)techniques were used to test wheat flour.The main results are as follows:1.Quantitative analysis of wheat flour quality was carried out by partial least squares(PLS)and support vector machine(SVM).In the quantitative analysis of wheat flour protein,wet gluten and ash infrared based on photoacoustic spectroscopy(FTIR-PAS),the root mean square error(RMSEP)of the optimal prediction models were 0.362%,0.229%and 0.019%,and the RPD values were 2.21,4.67 and 2.74,respectively.In the quantitative analysis of wheat protein,wet gluten and ash based on mid-infrared spectroscopy,the root mean square error(RMSEP)of the optimal prediction models were 0.299%,3.332%and 0.023%respectively,and the RPD values were 3.47,2.33 and 2.25,respectively.In the quantitative analysis of wheat protein,wet gluten and ash based on near-infrared spectroscopy(NIR),the root mean square error(RMSEP)of the optimal prediction models were 0.151%,0.516%and 0.041%respectively,and the RPD values were 6.05,3.88 and 3.44,respectively.2.For the distinction of wheat varieties,the characteristic spectra,principal component analysis(PCA),significant difference analysis of physical and chemical indexes and different discriminant models were analyzed respectively.In the characteristic spectral analysis,the spectral differences among different wheat cultivars can be clearly observed.Among them,the spectral differences between 1 800-800 cm-1 and 3 700-2 500 cm-1 are obvious,and the N-H and C-O bonds in protein are concentrated in this region,this laid the foundation for qualitative analysis.Based on principal component analysis,it was found that based on wheat infrared photoacoustic spectroscopy,PC1 can explain the total contribution variable of 81.56%,and PC2 can explain 12.88%.’Shannong 20’ and ’Heijin 2’can be better distinguished.;based on the mid-infrared spectroscopy,PC1 can account for 85.05%of the total contribution variable,and PC2 can account for 7.35%.’Jiangmai 816’ and ’Zhongmai 895’ cannot be distinguished,and can be distinguished from ’Heijin 2’;based on near-infrared spectroscopy,PC1 can account for 93.24%of the total contribution variable,and PC2 can explain 4.9%,indicating that all four wheat types can be effectively classified.By principal component analysis,the distinction of different types of wheat can be clearly observed.The physicochemical values of protein,wet gluten and ash of four wheat varieties were analyzed by single factor analysis and multiple Duncan comparison.Based on protein analysis,the content of ’Zhongmai 895’ was the highest and ’Heijin 2’was the lowest.Based on the analysis of wet gluten,’Zhongmai 895’ had the highest content and ’Heijin 2’ had the lowest.Based on the analysis of ash content,’Shannong 20’had the lowest ash content,while ’Heijin 2’,’Zhongmai 895’ and ’Jiangmai 816’ had the same ash content.There were significant differences in physical and chemical indexes among different wheat flour varieties.3.Partial least squares discrimination(PLS-DA)and Support vector machine(SVM)were used to identify four different wheat flour varieties.Finally,based on SPA-SVM-DA model,the best discrimination results were obtained for distinguishing wheat varieties by three different spectral techniques.In the distinction of ’Heijin 2’,’Jiangmai 816’,’Shannong 20’ and ’Zhongmai 895’ based on infrared photoacoustic spectroscopy(FTIR-PA S),the correct rate of model set and prediction set of the best discriminant model was 93.9%and 93.3%,respectively.In the distinction of four varieties based on mid-infrared spectroscopy(MIR),97.2%of the model set and 91.7%of the prediction set were obtained.When four varieties were discriminated by near infrared spectroscopy(NIR),the correct rates of model set and prediction set of the best discriminant model were 97.8%and 96.7%respectively.According to the results,it can be seen that the wheat flour varieties discrimination based on infrared photoacoustic spectroscopy is feasible,and the SPA-SVM-DA model is superior to the SPA-PLS-DA model.