近红外反射光谱技术快速测定小麦中必需氨基酸含量的研究

Study on Rapid Determination of Essential Amino Acid Contents in Wheat by Near-Infrared Reflectance Spectroscopy(NIRS)

试验探讨了近红外反射光谱快速测定小麦中10种必需氨基酸含量的可行性，并考察了不同光谱预处理方法和波长范围对近红外定标效果的影响。从全国范围内收集450个不同品种和产地的小麦样品．使用改进偏最小二乘法，采取77种导数和去散射光谱预处理方法建立近红外定标模型。结果表明：对于大多数氨基酸。二阶导数产生了较好的预测效果；各氨基酸定标模型建立的最佳波长范围并不一致。除蛋氨酸和色氨酸外，其余氨基酸均取得较好定标效果，其RSQd、（1-VR）和RPD。分别为0．84—0．94、0．77—0．923和2．09～3．68，且这些定标模型具有较好的预测能力，其RSQv和RPDv分别为0．77-0．92和2．14～3．64。除色氨酸外，近红外光谱法对小麦中必需氨基酸的预测效果优于粗蛋白回归法。

The possibility of using near-infrared refl ectance essential amino acid contents in wheat and the influence of methods of spectrum on the NIR calibration were investigated in spectroscopy （NIRS） for rapid determination of 10 different spectral ranges and different pre-treatment the present study. A total of 450 wheat samples with different varieties and planting locations were collected in China over a three years period. Calibrations were performed by modified partial least squares （MPLS） algorithm and 77 different derivatives plus scatter correction spectral pretreatments. The results showed that the second derivative mathematical treatment gave the best predictive performance for most amino acids. The best spectral range of different amino acid calibration model was not consistent. Except for methionine and tryptophan, relatively good calibration results were obtained for the remaining 8 amino acids, their RSQ~, （1-VR） and RPDcv ranged from 0.84 to 0.94, 0.77 to 0.93, and 2.09 to 3.68, respectively, and the calibration models for them also had relatively high predictive ability, their RSQv and RPDv ranged from 0.77 to 0.92 and 2.14 to 3.64, respectively. Except for tryptophan, compared to the linear regression results of amino acid contents relative to crude protein, NIRS has a better predictive performance.