应用便携式近红外光谱仪研究苹果糖度的快速分析模型

Rapid Analysis Model of Apple Sugar Degree Using Portable Near-Infrared Spectrometer

为了利用便携式近红外光谱仪实现苹果糖度快速无损的现场检测。利用K-S算法划分样本集,偏最小二乘法和间隔最小二乘法算法优选苹果糖度的最佳波长区间,对比分析9种预处理组合,建立苹果糖度的近红外预测模型。结果表明:使用偏最小二乘法选出优化糖度区间为1 198～670 cm^(-1),使用一阶求导、S-G卷积平滑和标准归一化处理后的模型预测性较好,校正集的相关系数和均方根系数分别为0.922 3和0.423,预测集的相关系数和均方根系数分别为0.918 9和0. 237。研究表明,利用精度较低的便携式光谱仪器可以快速、无损地现场检测出苹果糖度。

In order to realize the fast lossless field detection of apple sugar by portable near infrared spectrometer, the K- S algorithm was used to divide the sample set. The partial least square and the interval partial least square algorithm was used to determine the optimal wavelength range of apple sugar. After compared and analyzed 9 pretreatment combinations, apple sugar s near-infrared prediction model was set up. The results showed that the optimal sugar range was 1 198-670 cm -1 using the partial least square method, and the model with first-order derivation, S- G convolution smoothing, and standard normalization was the best predictor. The correlation coefficients and RMSEC of the correction set were 0.922 3 and 0.423, and the correlation coefficients and RMSEP of the prediction set were 0.918 9 and 0.237. The results showed that the apple sugar could be detected quickly and lossless by using the portable spectrometer with low accuracy.