摘要
应用自行搭建的CCD近红外光谱系统检测苹果的糖酸度。通过Y型光纤采集120个红富士苹果的漫反射光谱,采用偏最小二乘回归(PLSR)建立苹果糖度、酸度的定量预测模型。针对CCD光谱噪声较大的特点,采用S-G平滑、一阶导数、二阶导数对光谱进行预处理。结果表明,S-G平滑后所建模型的效果最好,糖度、酸度的相关系数(r)分别为0.9240、0.8151,标准校正误差(SEC)分别为0.9254、0.0120,标准预测误差(SEP)分别为0.9407、0.0204。本研究说明应用CCD近红外光谱仪,在630~1030nm波段实现对苹果糖度、酸度的无损检测具有可行性。
In this research, it was performed the measurement of apple's sugar content and acidity by using CCD-NIR spectroscopy system, with the Y-type fiber acquiring the reflectance spectra of 120 Fuji apples. Partial least squares regression was applied to construct the model between measured values and spectral signals. It was performed different methods of data pretreatment such as S-G smoothing, first derivative and second derivative in order to reduce the CCD spectrum noise. The results show that the best models of the sugar content and acidity are obtained after preprocessing of S-G smoothing. The calibration model gives the correlation coefficients of 0.9240 and 0.8151 respectively, with the standard error of calibration of 0.9254 and 0.0120 respectively, and the standard error of prediction of 0.9407 and 0.0204 respectively. The study showes that it is feasible to nondestructively measure the sugar content and acidity of apple using CCD-NIR system in 630-1030 nm.
出处
《食品科学》
EI
CAS
CSCD
北大核心
2007年第8期376-380,共5页
Food Science
关键词
苹果
糖度
酸度
近红外
CCD
光谱预处理
偏最小二乘法
apple sugar content acidity near infrared CCD spectral pretreatment partial least squares