| In this manuscript, the near infrared spectrum data has been collected from Malus asiatica samples during different shelf life combined with piercing quality indexes from texture experiments, then different chemometrics models have benn evaluated to precidition and classification of Malus asiatica samples during different shelf life. The main research contents and conclusions are as follows:(1) The TMS-PRO food property analyze has been used to puncture test the240samples of Malus asiatica. A new index has been defined as softening index, which according to the change of firmness and shelf life. The new index has been used to analysis the quality of Malus asiatica samples during different shelf life. The results showed that the softening index had a similar reduce trend as firmness during the extension fo shelf life. But compared with firmness statistical results, the softening index had the smaller standard deviation and variance, the redece trend was more flat and overlap phenomenon has been eliminated.(2) Firmness and softening index models have been extablished combining five different pretreatment methods (S-G、MSC、De-trending、SNV、Baseline) with three different models(PLS、PCR、 LS-SVM). The result showed that the best prediction of model has been built by PLS model, which softening index model had higher accuracy than firmness model. It was said that there had higher fitting phenomenon between the softening index with spectral data, the softening index can been used to classification the Malus asiatica. samples during different shelf life.(3) The nonlinear model(LS-SVM) and linear model(PLS-LDA)have been established to classification of the shaguo shelf life. The results indicated that non-linear LS-SVM model was more suitable for classification of Malus asiatica samples than linear PLS-LDA model, for the calibration and prediction set of LS-SVM model, the average correct recognition rate and average correct rejection rate above94%for both. |
PDF Full Text Request