食品科学 ›› 2018, Vol. 39 ›› Issue (14): 257-262.doi: 10.7506/spkx1002-6630-201814038

• 成分分析 • 上一篇    下一篇

斜率/截距算法在猕猴桃可溶性固形物含量便携式检测中的应用

李倩倩,刘大洋,杨彪,李伟强,郭文川*   

  1. (西北农林科技大学机械与电子工程学院,陕西?杨凌 712100)
  • 出版日期:2018-07-25 发布日期:2018-07-16
  • 基金资助:
    “十二五”国家科技支撑计划项目(2015BAD19B03-02)

Application of Slope/Bias Algorithm in Portable Detection of Contents of Soluble Solids of Kiwifruit

LI Qianqian, LIU Dayang, YANG Biao, LI Weiqiang, GUO Wenchuan*   

  1. (College of Mechanical and Electronic Engineering, Northwest A&F University, Yangling 712100, China)
  • Online:2018-07-25 Published:2018-07-16

摘要: 基于微型光谱仪搭建一台体积小、易操作的猕猴桃可溶性固形物含量(soluble solids content,SSC)检测装置,并用该装置采集了‘华优’、‘徐香’和‘西选’3?个品种猕猴桃的近红外光谱,利用不同特征波长提取方法从全光谱中提取特征波长,并比较不同方法提取的特征波长及全光谱对3?个品种猕猴桃SSC的偏最小二乘模型预测精度的影响;用斜率/截距算法结合‘华优’猕猴桃大样本模型,预测‘徐香’和‘西选’猕猴桃的SSC。结果表明,连续投影算法对于模型简化效果最好,其对‘华优’、‘徐香’和‘西选’猕猴桃大样本的预测均方根误差(root mean square error of prediction,RMSEP)分别为0.583、0.678?°Brix和0.646?°Brix;用斜率/截距算法对‘华优’猕猴桃SSC模型进行校正时,仅用10?个‘徐香’和50?个‘西选’猕猴桃便能有效地提高对SSC的预测性能,其RMSEP分别为0.966?°Brix和0.875?°Brix。本研究为进一步构建精度更高、更便捷的微型集成式猕猴桃SSC检测仪提供理论依据。

关键词: 猕猴桃, 可溶性固形物, 近红外光谱, 便携式, 斜率/截距算法

Abstract: To conveniently and nondestructively detect the soluble solids content (SSC) of kiwifruits, near infrared (NIR) spectra of ‘Huayou’, ‘Xuxiang’, and ‘Xixuan’ kiwifruits were obtained by using a portable system for SSC detection in kiwifruit based on a mini-spectrometer. The effective wavelengths were selected by different methods, and the effects of the selected characteristic wavelengths and the full spectra on the prediction precision of the models established by partial least square regression (PLSR) were compared. Slope/bias (S/B) algorithm was used to correct the model for ‘Huayou’ kiwifruits to predict the SSCs of ‘Xuxiang’ and ‘Xixuan’ kiwifruits. The results showed that the SPA algorithm could effectively extract the effective wavelengths. The root mean square error of predictions (RMSEPs) were 0.583, 0.678 and 0.646 °Brix for ‘Huayou’, ‘Xuxiang’, and ‘Xixuan’ kiwifruits, respectively. A total of 10 ‘Xuxiang’ and 50 ‘Xixuan’ samples were sufficient to improve the prediction performance of the corrected model with RMSEP of 0.966 and 0.875 °Brix, respectively. This study can offer useful information for developing a more accurate and portable miniature integrated detector for kiwifruit SSC.

Key words: kiwifruit, soluble solids content, near infrared spectroscopy, portable, slope/bias (S/B) algorithm

中图分类号: