| The experiment selected sweet persimmon and astringent persimmon as experimentalmaterial. To research calibration model which identification of persimmon andnon-destructive testing persimmon by Near-infrared diffuse reflectance spectroscopy, andto evaluate the applicability of NIR spectroscopy technique for nondestructivemeasurement of quality in persimmon after picking. Using different pretreatment methodsand different band selection, study the internal quality in different varieties of persimmonafter postharvest temperature storage, cold storage and storage shelves. In this paper, themain research content and results are as follows:(1)Established discriminant model of different varieties of persimmon, cross-validationcorrelation coefficient (RCV) was0.9451; cross-validation error (SECV) was0.1184, theprediction model to determine the correct rate of98%. The discriminant model of differentstorage of sweet persimmon, RCVwas0.8956，SECV was0.2074, its predictive model todetermine the correct rate reached88.9%-100%. The discriminant model of differentstorage of astringent persimmon, RCVwas0.8735，SECV was0.2103, its predictive modelto determine the correct rate reached93.3%. The results showed that the near-infrareddiffuse reflectance spectroscopy can be used to quickly and accurately identification ofvarieties and storage of the persimmon.(2)To establish calibration model of sweet persimmon for storage period at roomtemperature, at cold temperature and complex respectively by NIR nondestructive testingtechnology. Using optimal model to predict unknown samples, the results showed that themodel of sweet persimmon provided better prediction performance for L, a, b, peel brittleand cohesive, the correlation coefficient of prediction(RP2)of0.908,0.933,0.919,0.862,0.862respectively, ratio performance deviation(RPD)of3.17,3.67,3.36,2.63,2.56respectively. The model of sweet persimmon provided only a rough quantitative analysisfor SSC, peel strength, the average hardness of flesh firmness and chewiness, RP2of0.832,0.858,0.82,0.844,0.837respectively, RPD of2.416,2.47,2.35,2.40,2.37respectively.Thus, nondestructive testing of quality of sweet persimmon after picking using NIRspectroscopy technique was feasible.(3)To establish calibration model of astringent persimmom for storage period at roomtemperature, at cold temperature and complex respectively by NIR nondestructive testingtechnology. Using optimal model to predict unknown samples, the results showed that themodel of astringent persimmon provided better prediction performance for a, b, andhardness, the correlation coefficient of prediction(RP2)of0.927,0.913,0.911respectively, ratio performance deviation(RPD)of3.61,3.31,2.75respectively. The model of sweetpersimmon provided only a rough quantitative analysis for L, SSC, and tannic, RP2of0.868,0.831,0.734，respectively, RPD of2.49,2.15,2.11respectively. Thus, nondestructive testingof quality of astringent persimmon after picking using NIR spectroscopy technique wasfeasible.(4)Apply different pattern recognition technology to establish integrated calibrationmodel for room temperature strorage, cold strorage and after storage shelves of differentvarieties of persimmon. Study the feasibility for SSC, hardness and color, the choice of the1000calibration sets, analog line detection technology. The results showed that the modelof sweet persimmon provided better prediction performance forL, a, and b, the correlationcoefficient of prediction (RP2) of0.923,0.960,0.949respectively, ratio performancedeviation (RPD) of3.53,4.98,4.14respectively. The model of sweet persimmon providedonly a rough quantitative analysis for SSC and hardness, RP2of0.781,0.711respectively,RPD of1.83,1.69respectively.SSC, hardness, L, a, and b of the integrated calibration model of astringentpersimmom, RP2was0.616,0.796,0.712,0.780,0.802respectively, RPD was1.69,1.83,1.61,1.77,2.03respectively. Thus, a large number of samples as the calibration set, thepredictive power of the model needs to be further improved. |
PDF Full Text Request