以氣相層析圖及拉曼光譜圖之圖譜特徵在模式識別法下進行油品快篩的研究
No Thumbnail Available
Date
2016
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
以質譜技術與拉曼光譜為基礎，使用 LabVIEW (Laboratory Virtual Instrumentation Engineering Workbench) 電腦語言程式，以模式識別 (pattern recognition) 的方法，成功開發了識別相似度的程式，可用來研究油品安全及摻偽快速篩選的判讀。藉由計算標準樣品和參考樣品之間的交叉相關係數 (cross correlation factor , CCF) 來評估相似度的值。程式中定義兩組圖譜的相似度， A 圖譜為標準品， B 圖譜為對照圖，值會在 100% 到 0% 之間。當標準樣品的圖譜特徵和參考樣品相似時， CFF 的值會接近1。當計算相同的兩圖譜的 CFF 值時會得到1；相反的，當兩樣品的圖譜特徵完全不同時，會得到接近0的值。也就是說，當 (A⋂B)/(A⋃B)=1 時，相似度為 100% ；當 (A⋂B)/(A⋃B)=0 時，相似度為 0% 。 CFF 是由個人電腦和用 LabVIEW 所寫出來的程式所計算得到。並可將此程式應用到層析圖譜及拉曼光譜上。以 Florihana 野生高地真正薰衣草香精油的氣相層析圖為標準圖譜，比對市售6種香精油在同條件下的氣相層析圖，相似度在 30% 到 85% 之間。此外，以拉曼光譜測量各種市售橄欖油，使用相同程式進行相似度的評估。選擇日清純橄欖油作為標準品，和其餘4種市售橄欖油做比對。發現高價格樣品相似度較高，低價格相似度較低，但都在 80% 以上。而最貴的松露橄欖油則因為添加松露只得 60.56% 的相似度。自行開發的程式具有簡單操作、將相似度量化、快速比對圖形結果等優點，且已成功應用在層析圖譜及拉曼光譜的比對上。
A program based on pattern recognition of data, obtained by GC/MS (gas chromatogram/mass spectrometry) and Raman spectrometry, is employed for the rapid screening of oils and related commodities. The degree of similarity is evaluated quantitatively by calculating a cross correlation factor (CCF) between the standard sample and reference commodities. We defined the similarity between A-spectrum and B-spectrum, either obtained from GC/MS or Raman, in the range from 100 to 0 %. The CCF value is close to unity when the spectral feature of the standard is similar to that of the reference; CCF = 1 when CCF is calculated between the same spectrum. In contrast, the CCF value is close to zero when the spectral features are completely different from each other. In the other words, when (A⋂B)/(A⋃B) = 1, similarity is considered as 100%, whereas when (A⋂B)/(A⋃B) = 0, then similarity is counted to 0%. The CCF was calculated using a personal computer and the program was written in LabVIEW (Laboratory Virtual Instrumentation Engineering Workbench). As a result, when the Florihana lavender vera wild essential oil was selected as the standard sample, 6 types of commercial lavender oils were compared. We found that, based on the GC/MS data, the other commercial essential oils provide 30 ~ 85% similarities. On the other hand, based on the Raman data, when the Nissin olive oil was selected as the standard sample, 4 types of commercial olive oils were also compared. The findings show that the other commercial olive oils provide 60 ~ 95% similarities.
A program based on pattern recognition of data, obtained by GC/MS (gas chromatogram/mass spectrometry) and Raman spectrometry, is employed for the rapid screening of oils and related commodities. The degree of similarity is evaluated quantitatively by calculating a cross correlation factor (CCF) between the standard sample and reference commodities. We defined the similarity between A-spectrum and B-spectrum, either obtained from GC/MS or Raman, in the range from 100 to 0 %. The CCF value is close to unity when the spectral feature of the standard is similar to that of the reference; CCF = 1 when CCF is calculated between the same spectrum. In contrast, the CCF value is close to zero when the spectral features are completely different from each other. In the other words, when (A⋂B)/(A⋃B) = 1, similarity is considered as 100%, whereas when (A⋂B)/(A⋃B) = 0, then similarity is counted to 0%. The CCF was calculated using a personal computer and the program was written in LabVIEW (Laboratory Virtual Instrumentation Engineering Workbench). As a result, when the Florihana lavender vera wild essential oil was selected as the standard sample, 6 types of commercial lavender oils were compared. We found that, based on the GC/MS data, the other commercial essential oils provide 30 ~ 85% similarities. On the other hand, based on the Raman data, when the Nissin olive oil was selected as the standard sample, 4 types of commercial olive oils were also compared. The findings show that the other commercial olive oils provide 60 ~ 95% similarities.
Description
Keywords
模式識別, 氣相層析質譜儀, 拉曼光譜, LabVIEW, pattern recognition, GC/MS, Raman, LabVIEW