茶葉品質近紅外光譜非破壞性檢測之研究

Non-Destructive Evaluation of Tea Quality Using Near Infrared Spectroscopy

王慶茵(Ching-yin Wang)

指導教授 ： 陳世銘

國立臺灣大學/生物資源暨農學院/生物產業機電工程學研究所/碩士(2010年)

https://doi.org/10.6342/NTU.2010.10411

摘要

茶葉主要含有兒茶素類（catechins）與茶胺酸（theanine）兩種化學成份，會隨茶種與發酵程度不同，表現不同的香氣與味道。兒茶素是常用於茶製程中，是監測醱酵度之重要指標，而茶胺酸則會影響茶之甘味，兩者含量與茶葉內部品質有密切相關。傳統應用HPLC量測茶葉化學成份的方法，既費時且耗成本更無法應用於線上檢測；基於傳統檢測上的不方便及限制，並期能應用於製程中的快速檢測，本研究應用近紅外光檢測技術，建立茶品內部品質之非破壞性檢測模式。
實驗對象為台茶12號，並將未醱酵、部份醱酵、全醱酵三種製程所製之茶葉，以全波段（400∼2498 nm）利用修正型部份最小平方迴歸（modified partial least linear regression, MPLSR）以及多重線性迴歸（multiple linear regression, MLR）兩種方法，建立茶葉、茶粉、茶湯三種型態之茶品樣本的兒茶素類與茶胺酸含量之檢量模式，並尋找其特徵波長。另使用400∼1098 nm波段光譜模擬影像感測器之範圍，利用MPLSR及MLR方法分析，分析最佳預測兒茶素類含量之模式及特徵波長，以提供未來開發茶葉多光譜影像即時檢測系統之參考。
在兒茶素類檢量模式中，全波段（400∼2498 nm）之分析，以二次微分MPLSR茶葉型態樣本之結果為最佳：特徵波段有三組，各為500∼1098 nm、1300∼1400 nm、1600∼1700 nm，其相關係數rc及rv皆高達0.99、SEC為4.94 mg/g、SEV為5.14 mg/g。在400∼1098 nm波段之分析，以原始光譜MPLSR茶葉型態樣本之結果為最佳，特徵波段為700∼1000 nm，其相關係數rc及rv皆達0.98、SEC為8.29 mg/g、SEV為8.53 mg/g。而在400∼1098 nm原始光譜MLR分析茶葉型態光譜中，以762、900、986、992、998 nm五個波長具有較高的相關性，其相關係數rc及rv為0.98及0.97。分析結果顯示兒茶素類確實可使用茶葉型態樣本之光譜所建立之檢量模式，預測其成分含量。在全波段（400∼2498 nm）之茶胺酸檢量模式分析中，最佳結果以二次微分MPLSR之茶粉樣本的檢量模式具有較高預測能力，其相關係數rc及rv各為0.71及0.67，SEC為0.26 mg/g、SEV為0.26 mg/g。茶湯分析結果顯示2186 nm是預測茶胺酸含量之特徵波長。
綜合上論，本研究成功利用近紅外光光譜，建立茶品之兒茶素類與茶胺酸之檢量模式與特徵波長，可提供未來開發茶葉多光譜影像即時檢測系統之參考。

關鍵字

茶葉 ； 近紅外光 ； 兒茶素類 ； 茶胺酸

並列摘要

Tea consists of two major constituents, catechins and theanine, whose amounts depend on tea species and degree of fermentation. Catechins and theanine result in the smell and flavor of tea. Catechins are usually an essential indicator used for monitoring the degree of fermentation during the tea processing. Moreover, theanine affects the sweetness of tea. Both of ctechins and theanine are regarded as the internal quality of tea. The traditional methods such as HPLC (high performance liquid chromatography) for detecting constituents in tea are too time consuming and high cost to be employed in the real time and on-line system. Therefore, the non-destructive inspection of tea internal quality using near infrared spectroscopy was studied in this study to overcome the inconveniences and restrictions of the traditional methods, and expected to be applied to the rapid inspection during the processing.
The tea, Ta-Cha No.12, was selected as samples in this study. Tea samples were produced by three different processes including unfermented, partially fermented, and fully fermented treatments. The inspection models of three tea product types (leaf, powder, and solution) for the measurements of catechin and theanine were established using the multiple linear regression (MLR) and the modified partial least squares regression (MPLSR) t the range from 400 to 2498 nm. The characteristic waveband was surveyed in the models. Regarding the range from 400 to 1098 nm, as the imager sensing range, was also analyzed using the MPLSR and MLR to develop the inspection model and to find the characteristic wavelengths to serve as a reference for the future development on multiple spectral imaging inspection system for tea.
The catechins calibration model of the second derivative spectra MPLSR of tea leaf in the range from 400 to 2498 nm was the best, and the selected spectral bands were 500~1098 nm, 1300~1400 nm and 1600~1700 nm. The values of rc and rv were 0.99, and the values of SEC, SEV were 4.94 and 5.14 mg/g. The best calibration model from 400 to 1098 nm was in original spectra MPLSR of tea leaf, and the selected spectral bands were 700~1000 nm. The values of rc and rv were 0.98 and the values of SEC, SEV were 8.29 and 8.53 mg/g. In MLR tea calibration model within 400∼1098 nm, with five wavelengths of 762,900,986, 992, 998 nm were selected due to higher correlations, in which the values of rc and rv were 0.98 and 0.97. As a conclusion that is the catechin in the tea leaves can be predicted by the NIR calibration model. In the theanine calibration model of the second derivative spectra MLR possessed higher prediction ability to the tea powder samples. The rc and rv were 0.71 and 0.67, and the SEC, SEV were 0.26 and 0.26 mg/g. According to the results, 2186 nm is a characteristic wavelength to predict the theanine in the tea powder form samples.
Overall, the calibration models and the characteristic wavelengths for catechins and theanine of tea were successfully established using near infrared spectroscopy in this study, and it provides a reference for the development of the multi-spectral real time imaging system for tea.

並列關鍵字

Tea ； Near Infrared ； Cathechins ； Theanine

參考文獻

洪悅慈。2006。結合茶品單寧及胺基酸含量檢測之螢光流動注射分析系統。碩士論文。台北：國立台灣大學生物產業機電工程學系所。

陳世銘、張文宏、謝廣文。1998。果汁糖度檢測模式之研究。中華農業機械學刊 7（3）：41-60。台北：中華農機學會。

張嘉麟、陳世銘。1999。近紅外光與核磁共振應用於乳粉脂肪含量之檢測。中華農業機械學刊 8（3）：43-59。

蔡志成。1993。以近紅外光檢測幾丁類物質去乙醯率之研究。碩士論文。台北：國立台灣大學生物產業機電工程學系所。

羅聖傑。2009。茶葉醱酵度之高光譜影像檢測。碩士論文。台北：國立台灣大學生物產業機電工程學系所。