近红外光谱技术结合偏最小二乘法检测大曲糖化力

摘要

糖化力是大曲的一个重要理化指标,传统检测方法过程繁琐,检测效率低,不能进行大批量样品检测.采用近红外光谱结合偏最小二乘法建立的大曲糖化力的预测模型,具有检测高效、样品预处理简单等优点.该模型的扫描波段数为5000~10000 cm-1,建模集的回归系数r为0.9613,验证集的回归系数r为0.9528;建模集标准偏差(SEC)与验证集标准偏差(SEP)的比值为0.9967.随机选取50组样品验证模型的预测准确性,相对于传统检测方法,该方法的平均相对误差为5.27 %,与传统检测方法不存在显著性差异,可用于大曲糖化力的预测.%Saccharifying power is an important physiochemical index of Daqu.Traditional detection method is cumbersome,ineffec-tive,and can not be applied in mass samples detection.In this study,the detection model of Daqu saccharifying power had been estab-lished by near infrared spectroscopy coupled with partial least squares method,which had the advantages of high efficiency and sim-ple pretreatment of samples.The scanning band number of the model was 5000~10000 cm-1,the calibration set regression coefficient was 0.9613,and the validation set regression coefficient was 0.9528.The ratio of the calibration set standard deviation(SEC)to the validation set standard deviation(SEP)of the model was 0.9967.50 sample groups were randomly selected to verify the prediction ac-curacy of the model,compared with traditional detection method,the average relative error of this model was 5.27 % and there was no significant difference between them.Accordingly,this model could be used to predict the saccharifying power of Daqu.