【摘要】 应用可见-近红外漫反射光谱在570～1848nm光谱区域内,建立了软枣猕猴桃总酚定量数学模型。实验将贮藏分三个阶段(采收阶段,贮藏12d,贮藏24d)进行,通过对比分析不同建模方法得到软枣猕猴桃总酚定标模型。结果表明,应用偏最小二乘回归算法、一阶导数处理和反相多元离散校正处理所建软枣猕猴桃总酚定标模型的预测性能较好。定标集样本的交互验证相关系数(RCV)为0.8939,交互验证均方根误差(RMSECV)为11.6734mg/100g;预测集样本的相关系数(RP)为0.8627,预测均方根误差(RMSEP)为16.7690mg/100g。研究表明:可见/近红外漫反射光谱对软枣猕猴桃总酚的快速无损检测具有一定的可行性,但模型精度有待提高。更多还原

【Abstract】 The objective of this study was to establish a mathematical model for quantitative determination of total phenols of Actinidia arguta using the near-infrared diffuse reflectance(NIR) spectra in the wavelength range of 570–1848 nm.The experiment was divided into three stages to set the model(harvest stage,storage of 12 days,storage of 24 days),and compared with the different modeling methods.The results showed that the partial least squares(PLS) model,with respect to the first order derivatives spectrum D1lg(1/R) and inverse multiple scatter correction(IMSC)treatment,provided better predictive performance for total phenols in Actinidia arguta.In calibration sample set,the correlation coefficient of cross validation(RCV) was 0.8939,and root-mean-square error of cross-validation(RMSECV) was 11.6734 mg/100 g.In prediction set sample,the correlation coefficient of prediction(RP) was 0.8627,and root-mean-square error of prediction(RMSEP) was 16.7690 mg/100 g.These results confirmed that it is feasible to use the established VIS/NIR spectroscopy model for the nondestructive determination of the total phenols in Actinidia arguta.However,further investigation is needed to improve the precision of the model.更多还原