The Application of CLS Algorithm in the Quantitative Analysis of Lime in Wheat Flour by NIR Spectroscopy

Classical least square (CLS) algorithm is applied in this thesis to develop the raw spectra corrected peak area model, the raw spectra corrected peak height model, the 2nd derivative spectra corrected peak area model and the 2nd derivative spectra corrected peak height model by NIR spectra data of the wheat flour samples with lime added in. The result indicated that the correlation coefficients of the 4 models are 0.9321, 0.9483, -0.9491 and -0.9482 respectively; the result of F-test indicated that a remarkable correlation exists between the specified values of lime in wheat flour and the the raw spectra corrected peak areas / heights or the 2nd derivative spectra corrected peak areas / heights, which indicated that CLS algorithm has a certain potential application in the quantitative analysis of lime in wheat flour by NIR spectra data. Meanwhile, the result of F-test indicated that a very remarkable correlation exists between the estimated and specified values both the calibration set and external validation set of the 4 models. The limit of detection of the 4 models are 4.83 %, 4.14 %, 4.14 % and 4.17 % respectively, which will be suitable for the rapid quality screening for the wheat flour in the market and will be of great importance to the quality screening of wheat flour in the market, guarantee of the customers' health and the design and manufacturing of the special NIR spectrometer.