近红外光谱法快速检测甜菜糖度的模型优化

Model optimization on rapid detection of beet sugar content by near infrared spectroscopy

中文关键词:  近红外光谱法  甜菜  糖度  偏最小二乘法

英文关键词:near infrared spectroscopy  sugar beet  sugar content  partial least squares

中文摘要:

      目的 建立起近红外光谱技术关于甜菜糖度的最佳预测模型。方法 研究了Savitzky-Golay平滑处理、Savitzky-Golay导数、均值中心化、差分求导、净分析信号、去趋势校正、标准正态变量变换和多元散射校正等8种预处理方法的多方法联用处理进行光谱数据的预处理, 结合光谱波段优选, 建立甜菜糖度与近红外光谱的预测模型。结果 在进行模型的评价时, 以误差均方根(SEP)、校正标准误差(SEC)与交叉检验误差(SECV)作为评价指标。结论 发现经过光谱波段优选之后, 结合Savitzky-Golay平滑、Savitzky-Golay导数、去趋势校正及均值中心化进行光谱数据的预处理得到的模型效果最佳。

英文摘要:

      Objective To establish the optimal forecast model of beet sugar content based on near-infrared spectroscopy. Methods Eight kinds of methods as Savitzky-Golay smoothing, Savitzky-Golay derivative, mean centering, differential derivative, net analytic signal, to the tendency calibration, standard normal variable transformations and multiplicative scatter correction were associated and combined with pre-ferable spectral bands choice to build the beet sugar content and the near-infrared spectrum prediction model. Results SEP, SEC and SECV were set as the evaluation index when evaluating the models. Conclusion The results show after the preferred spectrum bands choice, the preprocessing that combines Savitzky-Golay smoothing, Savitzky-Golay derivative, to the trend correction and mean centering can obtain the optimal model.