[Objective]A method for the detection of A2β-casein in milk based on mid-infrared spectroscopy is investigated.[Method]The MIR spectrum of milk samples were measured and preprocessed with multiple scattering correction and standard normalization to improve the signal-to-noise ratio.Then,the characteristic variables were selected by the competitive adaptive reweighting algorithm.The quantitative analysis model based on partial least squares was established.[Result]After pre-processing and variable screening,the training and prediction errors reached 0.061 g/100 g and 0.068 g/100 g,respectively,and the R reached 0.893 and 0.875,respectively.[Conclusion]The mid-infrared spectroscopy method can be used for the rapid detection of A 2β-casein in milk.The spectral pre-processing and characteristic variables extraction methods can significantly improve the modelling accuracy.It also provided a good basis for the spectral detection methods for other ingredients in milk.
万方数据