基于近红外光谱技术的马铃薯全粉蛋白质无损检测

Potato granules protein nondestructive testing based on NIR spectroscopy analysis technology

为了探索一种简捷、高效、快速、无损的马铃薯全粉品质的检测方法,近红外光谱技术被应用到马铃薯全粉蛋白质的无损检测研究中。以120个经过真空微波冷冻干燥技术处理的不同品种马铃薯为样品,进行光谱采集及相应的化学值测定,采用多元散射校正和SavitzkyGolay卷积平滑对1200-2400 nm波段范围内的原始光谱进行预处理,选取最优的预处理方法;运用主成分回归分析法与偏最小二乘回归系数分别选择特征波长,并根据蛋白质基团的特征光谱区间选取特征波段,建立全波段、特征波段和特征波长下的主成分回归和偏最小二乘蛋白质预测模型。结果为：经过多元散射校正处理后的光谱建模效果最好,且运用偏最小二乘回归系数选择特征波长建立的马铃薯全粉蛋白质校正和验证模型的相关系数和均方根误差分别为0.9693、0.2937和0.9779、0.3304,优于全波段和特征波段建立的模型。研究表明,采用近红外光谱技术对马铃薯全粉蛋白质的无损检测是可行的。

In order to explore a simple, efficient, rapid and Iossless method to test the quality of potato granules, NIR spectroscopy analysis technology can be applied to potato granules protein nondestructive testing. This paper selects 120 potatoes of different varieties after vacuum microwave freeze-drying treatment as the sample and carries out spectral acquisition and corresponding chemical value determination on the sample. Multiplicative scatter correction and Savitzky-Golay are used to pre process the original spectra whose band range lying between 1200-2400 nm. And the optimum preprocessing method is selected. Principal component regression and partial least-squares regression analysis are applied to select characteristic wavelength respectively and choose the characteristic bands according to the characteristic spectral range of protein group. The principal component regression model andpartial least squares prediction model under full wavelength, characteristic bands and characteristic wavelength are established. The results can be concluded as follows the spectra after multiple scatter correct treatment features the best modelling effect; the model of potato granules protein correction and verification using partial least squares regression coefficients under characteristic wavelength gives the correlation coefficient at 0.9693, 0.2937 and root-mean-square error at 0.9779, 0.3304, which are superior than the data obtained by the models under full wavelength and characteristic bands. The research just indicates that it is feasible to apply NIR spectroscopy analysis technology to potato granules protein nondestructive testing.