基于可见/近红外光谱预测枇杷糖度及模型优化

The establishment and optimization of the model for predicting the sugar content of loquat by Vis/NIR spectroscopy

为实现枇杷糖度的快速无损检测,并探究开阳枇杷糖度最优预测模型。首先利用光纤光谱仪获取开阳枇杷的反射光谱,分析比较标准正态变换和多元散射校正方法对原始光谱数据的预处理效果;然后基于原始全光谱和预处理后的全光谱数据分别构建预测开阳枇杷糖度的偏最小二乘回归和主成分回归模型;最后,采用连续投影算法和竞争性自适应重加权算法(competitive adaptive reweighted sampling,CARS)筛选特征光谱,并基于特征光谱构建预测开阳枇杷糖度的多元线性回归(multi linear regression,MLR)模型。结果表明,采用CARS算法从785个全光谱中筛选了23个特征波长,不仅提升了预测模型的运算效率,而且建立的CARS-MLR模型具有最佳的校正性能(R_(C)=0.89,RMSEC=0.62)和预测性能(R_(P)=0.89,RMSEP=0.65,RPD=2.29)。这表明利用可见/近红外光谱技术结合化学计量学预测开阳枇杷糖度是可行的,且CARS-MLR模型相对最优,为枇杷品质的无损快检和分选提供了理论依据与技术基础。

The optimum model for rapidly nondestructive predicting the sugar content of Kaiyang loquat was explored and established.The fiber-optic spectrometer was used to collect reflectance spectra of Kaiyang loquat.The preprocessing effectiveness of standard normal variation(SNV)and multi-scatter calibration(MSC)on the original spectra data was compared and evaluated.Furthermore,the partial least square regression(PLSR)and principal component regression(PCR)models were established based on original full spectra and preprocessed full spectra to predict the sugar content of Kaiyang loquat,respectively.Finally,the successive projection algorithm(SPA)and competitive adaptive reweighted sampling(CARS)were applied to select characteristic spectra.And the multi linear regression(MLR)model was established based on characteristic spectra to predict the sugar content of Kaiyang loquat.The results showed that 23 characteristic wavelengths were extracted by CARS algorithm from 785 full spectra.The working efficiency of the prediction model was not only improved,but also CARS-MLR model showed the best calibration ability(R_(C)=0.89,RMSEC=0.62)and prediction ability(R_(P)=0.89,RMSEP=0.65,RPD=2.29).Consequently,Kaiyang loquat by Vis/NIR spectroscopy and chemometrics could be used to predict the sugar content,and the CARS-MLR model was best.These results can provide important theoretical and technical basis for the rapidly nondestructive prediction and sorting the quality of loquat.