Research On Rapid Detection Of Soy Sauce Industrial Production Based On Near-infrared Spectroscopy

The industrial production of high-salt liquid-state fermentation soy sauce has gradually become large-scale,mechanized,and automated.Traditional chemical methods cannot meet the determination needs of industrial production of high-salt liquid-state fermentation soy sauce due to being time-consuming and labor-intensive,and the production model that relies on the experience of workers has been unable to adapt to modern production.Near-infrared spectroscopy can obtain the characteristic information of hydrogen-containing groups in samples.It is widely used in quality inspection and process control in the food industry because of its fast,non-destructive and easy-to-use online characteristics.In this study,near-infrared spectroscopy was used to establish a rapid detection model to predict the amino acid nitrogen of soy sauce,detect the endpoint of soy sauce blending and stirring,and assist in the stability test of soy sauce products.Provide technical support for near-infrared spectroscopy for soy sauce quality control and modern production.A rapid detection model of amino acid nitrogen was established.Firstly,four groups of abnormal samples were eliminated by Monte-Carlo sampling,and the SPXY algorithm was used to divide the calibration set and the validation set.Then,the influence of six common preprocessing methods on the model accuracy was compared,and the results showed that Savitzky-Golay smoothing was the optimal preprocessing method.Then,competitive adaptive reweighted sampling(CARS)and successive projection algorithm(SPA)are used to extract characteristic wavelengths and establish partial least squares(PLS)model and support vector regression(SVMR)model,respectively.It is found that the 26 characteristic wavelengths extracted by CARS are better than the 13 characteristic wavelengths and the full spectrum extracted by SPA.Finally,the SGS-CARS-SVMR model was established to predict the amino acid nitrogen of soy sauce in industrial production.The coefficient of determination of the validation set was 0.9860,the root mean square error of prediction was 0.0198,and the residual prediction deviation was 11.05.It was detecting the endpoint of soy sauce blending.First,ten near-infrared spectra were collected from the same batch of soy sauce,and it was found that the soy sauce with the solid additives completely dissolved and evenly mixed had similar spectra.In the process of s simulated production in the laboratory,observe the changes in the near-infrared spectrum,compare the results of principal component analysis(PCA),and calculate the moving block standard deviation(MBSD).It can be seen from the mutual verification of the three that nearinfrared spectra combined with MBSD can characterize the dissolution process of solid additives in soy sauce.In the industrial blending production sampling for verification,the nearinfrared spectra changes and PCA results show that near-infrared spectra can characterize the industrial blending production of soy sauce.Combined with the experience of engineers and laboratory determination results,the threshold value of MBSD is set to 0.01 to detect the endpoint of blending in the industrial production of soy sauce.Auxiliary product stability testing.Firstly,a rapid identification model of soy sauce varieties is established to prevent transportation errors or packaging errors,and the model accuracy of soft independent modeling of class analogy,linear discriminant analysis,and Support Vector Machines(SVM)algorithms are compared.As a result,the SVM is the best,and the prediction accuracy of the validation set is 97.14%.Then,the conformity test model was established with the conformity index as the standard,the conformity index of 30 standard samples was calculated,the conformity index threshold was set to 3,and the external verification accuracy rate was 100%.Preliminary exploration on fermentation law of high-salt liquid-state fermentation soy sauce.First,the physical and chemical indicators in the fermentation process of soy sauce were detected,and the changing laws of physical and chemical indicators were summarized.By correlation analysis and cluster analysis,it was found that the fermentation process of high-salt liquid-state fermentation soy sauce can be roughly divided into four stages.