Modeling of functional components of wheat bran based on near-infrared spectroscopy
In order to realize the non-destructive and rapid detection of the functional components of wheat bran,using Near infrared spectruminstrument(NIR)combined with Partial least squares regression(PLS),comparing the effects of different pre-processing method,different spectral range and number of principal factors on the modeling results,and screening out the best modeling condi-tions.When the pre-processing method for the detection model of moisture,ash,and fat was S-G derivative(Savitzky-Golay),the spectral range were 2020 nm～2230 nm,1000 nm～2500 nm,and 1000 nm～2500 nm,and the number of principal factors were 11,7,and 6,the model had the best performance.When the pre-processing method was Multiple scattering correction(MSC),the spec-tral range was 1860 nm～2020 nm,and the number of principal factors was 6,the detection model of protien had the best perform-ance.When the pre-processing method was S-G smoothing,the spectral range was 1100 nm～1350 nm,and the number of principal factors was 8,the detection model of dietary fiber had the best performance.The Correlation coefficient(RC)and Correlation coeffi-cientsof prediction(RP)were above 0.85,Standard error of cross validation(SEC V)and Standard error of prediction set(SEP)were lower than other conditions,the three values are close to each other on the best modeling conditions,satisfying optimal modeling principles.Validation of the best model using samples from the prediction set that were not involved in the modeling yielded the re-sults that the absolute deviations of the predicted values from the true values(measured by the national standard method)were all less than 5％,indicating that the model can be used for the detection of the five functional constituents of wheat bran.
near infrared spectroscopy(NIR)wheat branfunctional compositionmodeling
万方数据
维普
