基于红外光谱技术的葡萄酒模式识别
首发时间:2018-03-05
摘要:本实验利用红外光谱技术结合模式识别方法,建立了一种葡萄酒产地快速检测的方法。采集葡萄酒的近红外和中红外光谱,对两种光谱分别预处理后进行主成分分析(PCA),比较基于近红外光谱和中红外光谱建立的模型效果,得出使用近红外光谱建模效果更优。对模型进一步优化后,训练集样本识别率都达到100%。除产地为法国的模型外,其余模型拒绝率均大于85%。利用预测集样本对模型预测能力进行评价。除智利外其他5个产地模型样品识别率为100%,而大部分产地的拒绝率都大于85%,说明模型预测效果良好,能有效区分不同产地的葡萄酒。
For information in English, please click here
The Pattern Recognition of Wine Based on Infrared Spectroscopy
Abstract:In this study, a method of rapid detection of wine origins was established by using infrared spectroscopy combined with pattern recognition. Near infrared and mid-infrared spectra of wine were pretreated for principal component analysis (PCA), respectively. The results of near-infrared and mid-infrared spectra were compared. The results showed that model based on NIR spectra was more effective. After further optimized, recognition rate of training sample was 100%. Except the model based on France, the rejection rate of remaining models was greater than 85%. Prediction set was used to evaluate predictive ability of model. The recognition rate of the other 5 origins for prediction set except Chile was 100%, while the rejection rate of most origins is more than 85%, which shows that the model is relatively better in prediction and can distinguish different origins of wines effectively.
Keywords: wine pattern recognition infrared spectroscopy traceability
引用
No.****
同行评议
共计0人参与
勘误表
基于红外光谱技术的葡萄酒模式识别
评论
全部评论0/1000