【摘要】 【目的】为了实现苹果可溶性固形物含量（SSC）的精准、快速检测，推动苹果产业高质量发展。【方法】采用可见/近红外光谱技术，测定苹果样本的可溶性固形物含量与光谱数据，利用MSC、SNV、SG 3种预处理方法结合偏最小二乘法（PLS）、支持向量回归（SVR）算法构建预测模型，并对不同预处理与建模方案进行对比评估。【结果】采用MSC和SNV预处理的光谱数据具有较高的预测能力；MSC-SVR模型性能最优，其预测集决定系数达到89.55%，均方根误差达到48.88%，能够很好地实现对苹果可溶性固形物含量的无损检测。【结论】研发的无损检测装置是可行的。更多还原

【Abstract】 [Objective]To achieve precise and rapid detection of the soluble solid content(SSC) of apples and promote the high-quality development of the apple industry. [Method]The visible/near-infrared spectroscopy technique was used to measure the soluble solid content of apple samples and their spectral data. Three preprocessing methods(MSC, SNV, and SG) were combined with partial least squares(PLS) and support vector regression(SVR) to construct a prediction model, and different preprocessing and modeling schemes were compared and evaluated. [Result] The spectral data preprocessed by MSC and SNV had higher predictive ability; the MSC-SVR model performed the best, with a determination coefficient R~2 of the prediction set reaching 89.55% and a root mean square error RMSE of 48.88%, which could effectively realize non-destructive detection of apple soluble solid content. [Conclusion] The developed non-destructive detection device is feasible.更多还原