Study On The Stability And Optimization Of Soluble Solids Content Detection Model Of Pear Based On Near Infrared Spectroscopy

With the development of social economy and the improvement of people's living standard,the demand for fruit quantity is increasing and the requirement for fruit quality is also getting higher and higher.In addition,China's fruit planting area and yield are second to none in the world,but the exports are very small,and lack competitiveness in the international market.The main reason lies in the backward of post harvest commercialization processing technology of fruits.Therefore,enhance the post harvest industrialization and commercialization processing of fruits,which can not only satisfy people's high requirements for fruit quality,but also improve the competitiveness of China's fruit industry.In this paper,the soluble solids content(SSC)of pear from China was taken as the detection index(also called sugar content),using near infrared spectroscopy technology,combined with chemometrics methods,the influence of different harvest years,different storage time,and radial different detection positions on the stability of SSC detection model of pear was studied,the modeling analysis and compensation model was studied to achieve model optimization.The main research contents and conclusions of this paper were as follows:1.The influence of different harvest years on the stability of SSC detection model of‘Xue'pear was studied,and a method of extracting effective variables from the spectra of pears in different harvest years was proposed.‘Xue'pear samples collected in 2017,2018 and 2019 respectively were used as research objects,single harvest year models and hybrid harvest year model of‘Xue'pear SSC were established and compared.The results indicated that the hybrid harvest year model,which established by mixing all the calibration set samples of 3 years,had better prediction results for the prediction set samples of each year.Combined with the 22 effective variables selected from the hybrid model,the model was greatly simplified while ensuring prediction accuracy,the correlation coefficient for prediction(R_p)and root mean square error of prediction(RMSEP)for the prediction set samples of each year were 0.947 and 0.281°Brix,0.964 and 0.203°Brix,0.963 and 0.227°Brix,respectively.The prediction results for all the prediction set samples of 3 years were R_p=0.971,RMSEP=0.285°Brix.The hybrid model was external validated by the same‘Xue'pear samples purchased in January 2020,with R_p and RMSEP of 0.941 and 0.358°Brix.Based on the effective variables,the stability of the hybrid harvest year model was better,can achieve accurate prediction of‘Xue'pear SSC in different harvest years.This study provided a reference for reducing the influence of different harvest years on the stability of SSC detection model of pear.2.The influence of different storage time on the stability of SSC detection model of‘Cuiguan'pear was studied,and a method of extracting effective variables from the spectra of pears with different storage time was proposed.The same batch of‘Cuiguan'pear samples were divided into 4 groups,and stored for 7 days,14 days,21 days and 28 days respectively for spectra data collection and SSC measurement.Single storage time models and hybrid storage time model were established and compared,and the models were validated by the prediction set samples of each group.The results shown that the hybrid storage time model,which established based on all the calibration set samples of 4 groups,achieved better prediction results.Combined with the 27 effective variables selected from the hybrid model,the prediction error RMSEP of the model were reduced to 0.357,0.388,0.361 and0.349°Brix,respectively.The prediction results for all the prediction set samples of 4 groups were R_p=0.955,RMSEP=0.359°Brix.The hybrid model was external validated by the same batch of‘Cuiguan'pear samples stored for 10 days,with R_p and RMSEP of 0.939 and 0.322°Brix.The hybrid storage time model based on the effective variables can realize accurate prediction of‘Cuiguan'pear SSC with different storage time,the stability and generalization ability of the model were improved.This study provided a reference for reducing the influence of different storage time on the stability of SSC detection model of pear.3.The influence of radial different detection positions on the stability of SSC detection model of‘Cuiguan'pear was studied,and a method of extracting effective variables from the spectra of radial different detection positions of pears was proposed.The spectra data and SSC around the radial stem,equator and calyx of‘Cuiguan'pear samples were obtained respectively.Local detection position models and global detection position model were established respectively,and the prediction sets of each position were predicted by using these models.The results indicated that the global detection position model which contained the information of all the 3 positions had better prediction results.Combined with the 36 effective variables extracted from the global model,the RMSEP of the model were reduced to 0.355,0.324 and 0.360°Brix,respectively.The prediction results for all the prediction sets of 3 positions were R_p=0.936,RMSEP=0.350°Brix.The global model was external validated by the same‘Cuiguan'pear samples,with R_p and RMSEP of 0.935 and 0.383°Brix.Based on the selected effective variables,the stability of the global detection position model was better,and less affected by the change of the radial detection position of sample.This study provided a reference for accurate prediction of SSC at radial different positions of pear.