Rapid Determination Of Five Flavor Substances In Dukang Base Liquor By Near Infrared Spectroscopy

In the daily production of liquor,the detection of the physical and chemical indicators of liquor is essential.An important item in the physical and chemical test is the detection and analysis of flavor substances in base liquor.The content of flavor substances has an important influence on the quality and grade of base liquor.At present,the liquor flavor substance detection method recognized by the liquor industry is determined by gas chromatography(GC)with reference to GB/T10345-2007"Liquor Analysis Method".However,this method is tedious and time-consuming,and is not suitable for online detection and analysis of base liquor.Near infrared spectroscopy technology has been widely used in many industries due to its fast,convenient,efficient,environmentally friendly,non-destructive analysis and other characteristics.The application of near infrared spectral analysis technology to the rapid detection of liquor flavor substances will play an important role in the intelligent production of liquor.Base liquor produced by Dukang Winery in Luoyang was selected as the research object,the chemical values of 5 typical flavor substances including ethyl octanoate,?-phenylethanol,3-methyl butanal,2,3-butanedione and 3-hydroxy-2-butanone were detected by gas chromatography technology,and using Fourier transform near infrared spectroscopy instruments to collect sample spectral data.The quantitative analysis model was established by chemometric methods to complete the rapid detection of these five flavor substances in liquor.1.Use gas chromatography analysis technology for detection and optimize the most suitable chromatographic conditions through experiments:The temperatures of the injector and detector were set at 220 and 250°C respectively.The column velocity was 2.0m L/min;The carrier gas was nitrogen maintained at a flow rate of 30 m L/min.The air flow rate is 300 m L/min and H₂ flow rate is 30 m L/min.Split injection method is adopted,and the injection volume was 1?L with a split ratio of 10:1;The chromatographic program was set at 75°C,held for 3 min,then raised to 210?at10?/min.The gas chromatography method for the detection of ethyl octanoate,?-phenylethanol,3-methyl butanal,2,3-butanedione and 3-hydroxy-2-butanone was established in base liquor.The contents of five flavor substances in base liquor were briefly analyzed.2.Applying Fourier transform near infrared spectrometer to complete the near-infrared spectral data collection of base liquor samples,and the optimal instrument parameters were obtained through experimental optimization:the spectral scanning range was 12000?4000cm^-1,the instrument resolution was 8 cm^-1,and the number of scans was 32 times.3.The near infrared quantitative calibration model of ethyl octanoate,?-phenethyl alcohol,3-methyl butanal,2,3-butanedione,and 3-hydroxy-2-butanone in base liquor was established by chemometric methods.Principal component analysis was performed on the sample spectral data to select a representative sample for modeling,and the component concentration measured by gas chromatography(GC)was used as the chemical reference value.The experimental results show that the model established by partial least squares(PLS)is better than the model established by principal component regression(PCR)The optimal spectral pretreatment method and spectral band of each flavor substance were selected.The calibration set samples were combined with internal cross-validation to establish a quantitative calibration model.The validation set samples were used for external verification.The final correction set determination coefficients(R²)of the five flavor substances were 0.9528,0.9502,0.9245,0.9474,and0.9486;the corrected root mean square error(RMSECV)were 4.64 mg/L,1.08 mg/L,2.47 mg/L,5.55 mg/L,and 2.70 mg/L;the validation set determination coefficients(R²)were 0.9648,0.9449,0.9409,0.9546,and 0.9576;the root mean square error(RMSEP)of the prediction were 3.87 mg/L,1.11 mg/L,2.12 mg/L,5.01 mg/L,and 2.37 mg/L.4.The precision test was performed on the established near infrared quantitative model.The relative standard deviations of the near infrared prediction values of the five component contents were in the range of 0.43 to 2.18%,and the average relative error from the true value was in the range of 0.07 to 0.38 mg/L.It showed that the model built with good precision and high accuracy,and can satisfy the needs of rapid detection of flavor substances in the production of liquor industry.