UV-Vis-SWNIR漫反射光谱结合偏最小二乘法测定蕨菜总多糖含量

Prediction of the total polysaccharides content of Pteridium aquilinum by UV-Vis-SWNIR diffuse reflectance spectroscopy in combination with a partial least squares method

目的建立UV-Vis-SWNIR漫反射光谱结合偏最小二乘法测定蕨菜中总多糖的含量。方法收集140批样品,从中随机选取20份作为完全外部验证集样品,用于验证模型的可靠性和准确性;其余80批样品为校正集,40批样品为检验集,用于筛选蕨菜总多糖的定量模型光谱预处理方法、主因子数、UV-Vis-SWNIR光谱特征波长。结果采用一阶导数法对原始光谱预处理,主因子数为7,结合随机青蛙算法筛选建模波长为925~1084、890~910、560~820、525~540、451~520、385~445、251~270 nm时,所得预测模型校正集R^(2)值为0.9617,RMSEC值为0.167%,检验集R^(2)值为0.9332,RMSEP值为0.208%,完全外部验证误差范围为-0.741%~0.853%。结论该方法快速可行,可用于蕨菜总多糖含量的预测。

AIM To establish the UV-Vis-SWNIR diffuse reflectance spectroscopy in combination with a partial least squares method to determine the content of total polysaccharides in Pteridium aquilinum.METHODS Among the overall 140 collected P.aquilinum sample groups,20 were randomly selected as the complete external validation set for the verification of the reliability and accuracy of the model,80 were for the calibration set,and the other 40 were for the validation set,i.e.,the screening of the spectral pretreatment method,the number of main factors and the characteristic wavelength of UV-Vis-SWNIR diffuse reflectance spectroscopy of the model for the quantitative determination of P.aquilinum total polysaccharides.RESULTS The first derivative method was used to preprocess the original spectrum with a principal factor of 7.The random frog algorithm was used to screen the modeling wavelengths of 925-1084,890-910,560-820,525-540,451-520,385-445,251-270 nm.The result ant prediction model corrected set had R^(2) value of 0.9617,RMSEC value of 0.167%,R^(2) value for the test set of 0.9332 and RMSEP value of 0.208%.The complete external validation error range was from-0.741%to 0.853%.CONCLUSION This fast and feasible method can be used to predict the total polysaccharides content in P.aquilinum.