The aim of this experiment was to establish a mid-infrared spectral prediction model for the rapid de-tection of putrescine in raw milk.Raw milk samples collected from 16 farms were stored at 4 ℃ for 0 to7 d.The content of putrescine in raw milk was detected by high performance liquid chromatography-mass spectrom-etry method,and the mid-infrared spectrum of raw milk was collected by FT 120 milk composition analyzer.The original mid-infrared spectrum was preprocessed by standard normal variate transformation(SNV),Savitsky-Golay(SG)smoothing,first-order derivative or second-order derivative to eliminate background noise,and then partial least squares regression(PLSR)method was used to construct the putrescine prediction model.The results showed that by comparing different mid-infrared spectral preprocessing methods,the SG smoothing+second-order derivative preprocessing method was finally chosen to preprocess the spectra to estab-lish a prediction model,the model has an determination coefficient(R2)of 0.995 9 and an root mean square error(RMSE)of 0.054 2 for the test set and an R2 of 0.811 2 and an RMSE of 0.368 4 for the validation set,providing good accuracy and stability.In conclusion,a rapid method for the determination of putrescine in raw milk is established based on mid-infrared spectroscopy,which can provide reference for the rapid detection of putrescinein raw milk.[Chinese Journal of Animal Nutrition,2024,36(8):5432-5440]
关键词
中红外光谱/生乳/腐胺/偏最小二乘回归
Key words
mid-infrared spectroscopy/raw milk/putrescine/partial least squares regression
维普
万方数据