Eighty milk powder samples which represented over 10 formulae of ingredients from 11 commercial brands were collected and a PDA type near-infrared spectrometer was used to obtain their diffusion reflectance spectra (1nm resolution) within the wavelengths of 1089~2219nm. The obtained spectra were pre-treated with standard normal variate correction (SNV), wavelet denoise and 1-order differentiation method. Through comparing the weighted distribution of the milk powder’s five principal ingredients at various wavelengths, different ranges of wavelength were selected to establish calibration models and to analyze their prediction accuracy. The results showed that spectrum information of milk powder’s protein and fat composition was mainly distributed within the wavelengths of 1100~1400nm and 1800~2200nm. It was shown that wavelet denoise was an excellent method for pre-processing spectra, which could significantly enhance the stability of calibration models and prediction accuracy. The present study reveals that it is feasible to determine the concentrations of protein and fat in milk powder of various origins with a PDA type near-infrared spectrometer.
參考文獻
相似文獻
引證文獻
引用本文
顏輝,陳斌,朱文靜.基于近紅外光譜分析法的奶粉品質快速檢[J].農業(yè)機械學報,2009,40(7):149-152. Determination Method of Milk Powder Quality by Near-infrared Spectroscopy[J]. Transactions of the Chinese Society for Agricultural Machinery,2009,40(7):149-152.