基于介电频谱的采后苹果可溶性固形物含量无损检测

Soluble Solids Content Detection of Postharvest Apples Based on Frequency Spectrum of Dielectric Parameters

摘要:

根据10~4500MHz间采后21周贮藏期间无损富士苹果介电参数的频谱特性，建立了苹果可溶性固形物含量的支持向量回归（SVR）预测模型和BP网络预测模型；并综合比较了采用原始频谱（FF）、主成分分析（PCA）和连续投影算法（SPA）优选频率对模型预测效果的影响。研究结果表明，PCA—SVR建模效果最好，其预测相关系数为0.883，均方根误差为0.552，PCA—BP的建模效果较PCA—SVR稍差。并且发现经SPA处理后的数据建立的模型，均方根误差普遍较小；经PCA处理后的数据建立的模型，预测相关系数普遍较高。

Abstract:

Based on frequency spectrum of permittivities from 10MHz to 4500MHz of intact postharvest Fuji apples during 21 weeks storage, BP network model and support vector regression (SVR) model were applied to predict SSC. Effects of the prediction models using full frequency (FF), principal component analysis (PCA) and successive projection algorithm (SPA) were compared and evaluated. The results showed that PCA—SVR gave the best result rather than PCA—BP and SPA—BP. The predicted correlation coefficient of PCA—SVR was 0.883 and the root mean square error (RMSE) was 0.552. The effect of PCA—BP was a little worse than PCA-SVR. The RMSE of the model established by SPA was generally smaller than by other methods, and the predicted correlation coefficient of the models established by PCA was generally higher. The research offered some useful technologies in developing nondestructive sensors for fruits’ soluble solids content based on frequency spectrum of dielectric parameters.