| In this paper, dielectric loss factor ε" spectra and relative dielectric constant ε’spectra of 300 long jujubes were measured with a network analyzer under 101 selected frequency points in the frequency range of 0.2～18 GHz. The prediction model of soluble solids content was researched using the long jujube’s dielectric loss factor ε" spectra and relative dielectric constant ε’ spectra. The effective information of the dielectric spectra was extracted by genetic algorithm (GA) and correlation coefficient method (CC). Prediction model of soluble solids content was established using partial least squares (PLS), principal components regression (PCR) and support vector machine (SVM). dielectric loss factor ε" spectra and relative dielectric constant ε’spectra of long jujubes were measured with LCR tester under 55 selected frequency points in the frequency range of 1～1000KHz. The effective principal component of the dielectric spectra was extracted by principal component analysis (PCA). Identification model of damaged jujubes, jujubes of different maturity, different jujube varieties was established using partial least squares discriminant analysis (PLS-DA), linear discriminant analysis (LDA) and support vector machine (SVM).The results were shown as follows:(1) The research on the prediction methods of LingWu long jujube’s soluble solids content with network analyzer indicated as follows:Both the frequency and the storage time had significant influence on the dielectric parameters of the long jujube; The modeling effect with the extracted effective information using GA and CC was better than that of original spectra modeling, the modeling effect with PCR was better than that with PLS and SVM. The optimal prediction models of soluble solids content based on dielectric loss factor ε" and relative dielectric constant ε’ spectra were GA-PCR and CC-PCR respectively. The effect of GA-PCR modeling with dielectric loss factor ε" was better than that of CC-PCR modeling with relative dielectric constant ε’, the correlation coefficients of calibration set and prediction set were 0.933 and 0.925 respectively, and the root mean square errors (RMSE) were 0.661 and 0.702.(2) The research on identification methods of damaged LingWu long jujubes with LCR tester indicated as follows. Both the frequency and damage had significant influence on the dielectric parameters of the long jujube. Comparing the three modeling methods of PLS-DA, LDA and SVM, the SVM model was optimal and its predicted identification accuracy rate was 100%, The modeling effects with relative dielectric constant ε’ was better than that of dielectric loss factor ε". Therefore, the optimal identification model of long jujube’s damage was SVM model based on relative dielectric constant ε’ spectra.(3) The research on identification methods of maturity LingWu long jujube with LCR tester indicated as follows. Both the frequency and maturity had significant influence on the dielectric parameters of the long jujube. Comparing the three modeling methods of PLS-DA, LDA and SVM, the SVM model was optimal and its predicted identification accuracy rate was 100%, The modeling effects with dielectric loss factor ε" was better than that of relative dielectric constant ε’. Therefore, the optimal identification model of long jujube’s maturity was SVM models based on dielectric loss factor ε" spectra.(4) The research on identification methods of varieties jujube fruit with LCR tester indicated as follows. Both the frequency and varieties had significant influence on the dielectric parameters of the jujube fruit. Comparing the three modeling methods of PLS-DA, LDA and SVM, the PLS-DA model was optimal and its predicted identification accuracy rate was 100%, The modeling effects with dielectric loss factor ε" was better than that of relative dielectric constant ε’. Therefore, the optimal identification model of jujube fruit’s varieties was PLS-DA models based on dielectric loss factor ε" spectra. |
PDF Full Text Request