茉莉花茶的品質(zhì)評定與價格判別

張俊, 龔淑英, 唐德松, 張穎彬, 陳美麗

(1.浙江大學(xué)茶葉研究所,杭州 310058;2.中華全國供銷合作總社杭州茶葉研究院,杭州 310016;3.中國農(nóng)業(yè)科學(xué)院茶葉研究所,杭州 310008;4.柳州市林業(yè)科學(xué)研究所,廣西 柳州 545300)

茉莉花茶的品質(zhì)評定與價格判別

張俊1,2, 龔淑英1*, 唐德松1*, 張穎彬1,3, 陳美麗1,4

(1.浙江大學(xué)茶葉研究所,杭州 310058;2.中華全國供銷合作總社杭州茶葉研究院,杭州 310016;3.中國農(nóng)業(yè)科學(xué)院茶葉研究所,杭州 310008;4.柳州市林業(yè)科學(xué)研究所,廣西 柳州 545300)

以茉莉花茶為研究對象,采用審評沖泡法和全量浸提法,通過因子分析和判別分析建立基于化學(xué)成分的品質(zhì)評定和價格區(qū)間判別模型。結(jié)果表明:通過因子分析,審評沖泡法和全量浸提法分別得到累積解釋變異量達91.825%和93.858%的5個因子,并均由此得到5個因子得分函數(shù)和1個綜合評定函數(shù);通過判別分析,審評沖泡法和全量浸提法均得到3條貝葉斯(Bayes)判別函數(shù),回代總正判率分別為85.7%和83.9%,交叉驗證總正判率分別為84.8%和81.3%。這說明2種方法的正確判別力較高,且審評沖泡法的判別效果更好。

茉莉花茶; 因子分析; 判別分析; 審評沖泡法; 全量浸提法

茉莉花茶是花茶中的主流產(chǎn)品,尤其受中國北方消費者歡迎。為規(guī)范市場,保護消費者合法權(quán)益,做好茉莉花茶的品質(zhì)控制尤為重要。目前,世界各國對茶葉品質(zhì)、等級、價格的確定依然主要依靠感官審評這一傳統(tǒng)方法。該方法對審評人員的經(jīng)驗、感官靈敏度等要求較高,且受個人喜好、身體狀況等主觀因素影響較大。因此,探索一種客觀、穩(wěn)定、可量化的品質(zhì)評定與判別方法非常必要。

因子分析是一種將眾多關(guān)系錯綜復(fù)雜的變量合并綜合成較少的幾個綜合變量(因子)的方法,以達到用較少幾個因子反映原始資料的大部分信息的目的。判別分析則是一種判別和分類的方法,通過建立判別函數(shù)對其他未知分類的數(shù)據(jù)進行類別歸屬,具有預(yù)測意義[1]。因子分析和判別分析都是統(tǒng)計學(xué)中的常用方法,并已在自然科學(xué)、醫(yī)學(xué)、經(jīng)濟學(xué)、社會學(xué)等多個領(lǐng)域被廣泛應(yīng)用[2]。楊紅等[3]采用因子分析研究了各生化成分在茶湯冷后渾組分中的權(quán)重,葉國注等[4]將貝葉斯(Bayes)逐步判別法應(yīng)用于綠茶板栗香的化學(xué)識別中;周利兵[5]以微量元素為變量采用因子分析對我國六大名茶進行了綜合評價;成浩等[6]基于化學(xué)指紋圖譜對扁形茶產(chǎn)地和王麗鴛等[7]基于多元化學(xué)指紋圖譜對武夷巖茶身份進行了判別研究;孫威江等[1]應(yīng)用因子分析對烏龍茶品質(zhì)進行了綜合評定并對烏龍茶產(chǎn)地、品種和等級進行了判別分析等。

以往對于茶葉品質(zhì)成分的研究均按照國家標準,采用全量浸提方法獲取茶湯,該方法可測定茶葉中各品質(zhì)成分占干物質(zhì)量的百分比,但與茶葉品質(zhì)評定方式及人們?nèi)粘ｏ嫴璺绞讲顒e較大,且操作起來費時費力。本研究引入審評沖泡法這一與我們?nèi)粘Ｔu定茶葉品質(zhì)一致的茶湯獲取方法,以茉莉花茶為研究對象,從審評沖泡法和全量浸提法2個角度出發(fā),對茉莉花茶13個生化指標以及由此衍生出的19個指標總共32個變量進行因子分析,并將價格從28元/kg到3 000元/kg的茉莉花茶分成3個區(qū)間,對3個價格區(qū)間進行判別分析,以期為茉莉花茶的品質(zhì)評定和價格判別提供參考,并比較2種茶湯獲取方法對于茶葉品質(zhì)評定與價格判別效果的影響。

1 材料與方法

1.1 實驗材料

從北京吳裕泰茶業(yè)股份有限公司、北京張一元茶葉有限責任公司、北京二商京華茶業(yè)有限公司、北京元長厚茶葉有限公司4個廠家及茶葉市場采購的價格從28元/kg到3 000元/kg不同檔次的茉莉花茶(共112個茶樣)用于化學(xué)成分測定。

1.2 實驗儀器與設(shè)備

手提式高速萬能粉碎機;電熱恒溫水浴鍋;循環(huán)水式多用真空泵;電子天平;電熱恒溫鼓風干燥箱;紫外可見分光光度計;高效液相色譜儀。

1.3 實驗方法

1.3.1 化學(xué)品質(zhì)成分測定方法 將每個茶樣按2種方法獲取茶湯,對其中的品質(zhì)成分進行測定。一種按照審評沖泡的方法,直接測定感官審評時喝到的茶湯中13個品質(zhì)成分的質(zhì)量濃度/(mg/mL),稱為審評沖泡法。另一種按照全量浸提的方法,測定茶樣中總共含有的這13個品質(zhì)成分占干質(zhì)量的百分比,稱為全量浸提法。13個品質(zhì)成分包括:茶多酚及兒茶素組成[沒食子兒茶素(gallocatechin,GC),表沒食子兒茶素(epigallocatechin,EGC),兒茶素(catechin,C),表兒茶素(epicatechin,EC),表沒食子兒茶素沒食子酸酯(epigallocatechin gallate,EGCG),沒食子兒茶素沒食子酸酯(gallocatechin gallate,GCG),表兒茶素沒食子酸酯(epicatechin gallate,ECG),兒茶素沒食子酸酯(catechin gallate,CG),游離氨基酸總量,咖啡堿,沒食子酸(gallic acid,GA),水浸出物]。

1)審評沖泡法:按照茉莉花茶感官審評方法,稱取3.0 g茶樣于審評杯中,加150 mL沸水沖泡5 min,將茶湯趁熱過濾,用于13個品質(zhì)成分的質(zhì)量濃度測定。2)全量浸提法:將茶樣磨碎,按照GB/T 8312—2002方法浸提茶湯,即稱取3.0 g磨碎茶粉于500-mL錐形瓶中,加450 mL沸水,在沸水浴中浸提45 min,每10 min搖1次,趁熱加壓過濾,濾液冷卻后加水定容至500 mL,所得供試液用于13個品質(zhì)成分的含量測定。

13個品質(zhì)成分測定方法:茶多酚按GB/T 8313—2002測定;游離氨基酸總量按GB/T 8314—2002測定;水浸出物按GB/T 8305—1987測定。兒茶素組成、咖啡堿、沒食子酸測定參照Liang等[8]的方法:將浸提得到的供試液過0.45 μm濾膜,用于高效液相色譜(high performance liquid chromatograph,HPLC)分析.具體分析條件:島津LC-20A型高效液相色譜儀;Agilent TC-C18(2)分析柱(4.6 mm×250 mm,5 μm),柱溫30 ℃,檢測波長280 nm,流速1.0 mL/min;流動相A為0.5%冰醋酸+3%乙腈+96.5%超純水,流動相B為0.5%冰醋酸+30%乙腈+69.5%超純水;洗脫梯度:前35 min內(nèi)B相從30%線性增加至85%,接著以30%B保持5 min,進樣量10 μL。

數(shù)據(jù)采用SPSS 16.0軟件中的Analyze模塊進行因子分析和判別分析。因子分析方法:采用KMO(Kaiser-Meyer-Olkin)法判斷數(shù)據(jù)是否適合因子分析,因子萃取采用主成分分析法,因子旋轉(zhuǎn)采用最大變異法,因子負荷量絕對值舍棄下限設(shè)定為0.5;判別分析方法:以價格為分組變量,以因子分析所得因子為自變量,自變量采用同步進入法進行判別。

2 結(jié)果與分析

2.1 茉莉花茶品質(zhì)的因子分析及綜合評定

對種子進行精選，剔出霉變、損傷粒，將選好的種子連曬2-3天，以提高出苗率和整齊度。播種量應(yīng)控制在2.5-3公斤/畝，根據(jù)品種特性進行酌情增減。播種深度3-5厘米，等行距種植的行距為60厘米左右，大壟行種植的大行距為90-110厘米，小行距為40厘米。典型高產(chǎn)經(jīng)驗表明，耐密型品種密度可在5000株/畝以上，大穗型品種密度可在4000-4500株/畝左右。出苗后及時查苗、補苗并及早間苗定苗。缺苗時可在同行或相鄰行就近留雙株，若缺苗太多則應(yīng)及早補苗。一般3葉期間苗，5-6片可見葉時定苗。為提高群體整齊度，應(yīng)去除弱苗、病苗、蟲苗，留壯苗、勻苗、齊苗。

112個茉莉花茶茶樣的審評沖泡茶湯中13個成分的質(zhì)量濃度以及由此衍生出的3個質(zhì)量濃度、16個比值總共32項指標的因子分析結(jié)果見表1。通過因子分析共萃取出5個特征值大于1的公共因子,從F1到F5解釋變異量依次減少,5個公共因子的累積解釋變異量達91.825%,即這5個因子可以反映原32個變量所攜帶的91.825%的信息量。共同性表示每個變量被因子解釋的方差估計量,越大說明變量能被因子說明的程度越高,信息丟失越少,數(shù)值太低說明該變量不適合做因子分析,可在分析中被排除。由表1可知,32個變量的共同性均在0.7以上,適合進行因子分析。

表1 茉莉花茶審評沖泡法主要生化成分的因子分析結(jié)果

Table 1 Factor analysis of the main biochemical compositions of jasmine-scented tea by brewing method for sensory analysis

變量Variable因子FactorF1F2F3F4F5共同性ExtractionX10.8100.962X20.8960.981X30.9300.959X40.8950.861X5-0.5860.7110.896X60.7600.742X70.8400.913X80.7080.760X90.8660.949X100.8810.920X110.7600.778X120.7960.878X130.8080.720X140.9020.976X150.9490.969X160.9770.990X170.691-0.6210.923X180.9530.992X19-0.7940.934X200.9600.994X210.7880.954X22-0.6920.6690.942X230.9510.955X240.8370.990X250.9460.990X26-0.8810.946X27-0.7280.5350.912X280.8640.979X29-0.9020.963X30-0.7120.721X310.9430.994X32-0.620-0.6650.944特征值Eigenvalue13.6276.3814.4743.5671.336解釋變異量Variance/%42.58419.94113.98011.1474.174累積解釋變異量Cumulativevariance/%42.58462.52576.5058.65291.825

5個公共因子中,F1在兒茶素(X16)、酯型兒茶素(X15)、水浸出物(X3)、簡單兒茶素(X14)、GA/兒茶素(X29)、EGCG(X10)、咖啡堿/兒茶素(X26)、EC(X9)、EGC(X7)、茶多酚(X1)、ECG(X12)、GCG(X11)、GC(X6)、咖啡堿/水浸出物(X27)、GA/咖啡堿(X30)、C(X8)、氨基酸/兒茶素(X22)、簡單兒茶素/酯型兒茶素(X17)、GA/氨基酸(X32)、GA(X5)上負荷量較高(按負荷量絕對值由大到小排列,其他公共因子也按此排列).這些變量主要反映了兒茶素和GA的信息,其中與兒茶素關(guān)系為正而與GA關(guān)系為負,暫且將其命名為兒茶素及GA(—)因子,它綜合了32項生化指標42.584%的信息量;F2在氨基酸/茶多酚(X20)、兒茶素/茶多酚(X18)、咖啡堿/茶多酚(X25)、GA/茶多酚(X31)、咖啡堿/(茶多酚+氨基酸)(X28)、氨基酸/(咖啡堿+茶多酚)(X24)、茶多酚/水浸出物(X19)上負荷量較高,這些變量主要反映了茶多酚的信息,但與茶多酚的關(guān)系為負,因而將其命名為茶多酚(—)因子,它綜合了32項生化指標19.941%的信息量;F3在氨基酸/水浸出物(X23)、氨基酸(X2)、氨基酸/咖啡堿(X21)、氨基酸/兒茶素(X22)、GA/氨基酸(X32)上負荷量較高,這些變量主要反映了有關(guān)氨基酸的信息,我們將其命名為氨基酸因子,它綜合了32項生化指標13.980%的信息量;F4在咖啡堿(X4)、GA(X5)、簡單兒茶素/酯型兒茶素(X17)、咖啡堿/水浸出物(X27)上負荷量較高,反映的信息不是很明確,暫且將其命名為咖啡堿因子,它綜合了32項生化指標11.147%的信息量;F5在CG(X13)上負荷量較高,將其命名為CG因子,它綜合了32項生化指標4.174%的信息量。

表2為因子得分系數(shù)矩陣,據(jù)此可以寫出每個公共因子與32個變量間的因子得分函數(shù)。根據(jù)表1中每個因子的解釋變異量及累積解釋變異量,又可寫出由公共因子構(gòu)成的綜合得分函數(shù)(表3)。將32個變量的標準化值代入因子得分函數(shù),即可得到各個公共因子的得分,然后再代入綜合評定函數(shù),即可按F值的大小對茉莉花茶進行綜合評定。

表2 茉莉花茶審評沖泡法主要生化成分的因子得分系數(shù)矩陣

Table 2 Factor scores of the main biochemical components of jasmine-scented tea by brewing method for sensory analysis

變量Variable因子FactorF1F2F3F4F5X10.082-0.0350.0100.126-0.023X20.0310.0230.2200.1070.019X30.0820.0420.0630.0790.044X40.0730.0360.0690.344-0.030X5-0.0060.0530.0680.2150.145X60.0380.027-0.027-0.0440.243X70.0540.0090.006-0.058-0.066

續(xù)表2 茉莉花茶審評沖泡法主要生化成分的因子得分系數(shù)矩陣

Continuation of Table 2 Factor scores of the main biochemical components of jasmine-scented tea by brewing method for sensory analysis

變量Variable因子FactorF1F2F3F4F5X80.0310.000-0.069-0.0590.241X90.032-0.009-0.015-0.0980.172X100.1280.0350.0350.215-0.242X110.0890.0360.0040.1620.107X120.062-0.005-0.0760.0230.135X13-0.0430.0190.073-0.0290.673X140.0470.004-0.011-0.0760.068X150.1150.024-0.0040.163-0.100X160.0810.014-0.0080.030-0.006X170.0070.0060.008-0.1740.099X180.0590.170-0.0440.054-0.050X190.030-0.112-0.0680.114-0.132X200.0290.1690.0610.052-0.006X21-0.010-0.0100.160-0.0810.011X22-0.086-0.0100.196-0.0020.180X23-0.0300.0000.2460.0680.062X240.0120.1450.1240.0340.014X250.0330.166-0.0400.074-0.022X26-0.067-0.021-0.0120.0230.085X27-0.028-0.020-0.0240.116-0.017X280.0190.150-0.0490.076-0.012X29-0.080-0.005-0.005-0.0090.179X30-0.0560.0560.0230.0180.204X310.0260.165-0.0390.0610.017X32-0.0200.019-0.1360.029-0.022

表3 茉莉花茶審評沖泡法品質(zhì)評定函數(shù)

Table 3 Quality evaluation functions of jasmine-scented tea by brewing method for sensory analysis

因子Factor函數(shù)Function兒茶素和GA(—)Catechin&GA(—)F1=0.082ZX1+0.031ZX2+0.082ZX3+ …+0.026ZX31-0.02ZX32茶多酚(—)Polyphenols(—)F2=-0.035ZX1+0.023ZX2+0.042ZX3+ …+0.165ZX31+0.019ZX32氨基酸AminoacidF3=0.01ZX1+0.22ZX2+0.063ZX3+ …-0.039ZX31-0.136ZX32咖啡堿CaffeineF4=0.126ZX1+0.107ZX2+0.079ZX3+ …+0.061ZX31+0.029ZX32CGF5=-0.023ZX1+0.019ZX2+0.044ZX3+ …+0.017ZX31-0.022ZX32綜合評定ComprehensiveevaluationF=(42.584F1+19.941F2+13.98F3+ 11.147F4+4.174F5)/91.825

GA：沒食子酸；CG：兒茶素沒食子酸酯；ZX：標準化變量值.

GA： Gallic acid； CG： Catechin gallate； ZX： Standardized variable values。

與審評沖泡法類似,對于112個茉莉花茶茶樣全量浸提后測得的13個成分占干質(zhì)量的百分比以及由此衍生出的3個含量、16個比值總共32項指標的因子分析結(jié)果見表4。

表4 茉莉花茶全量浸提法主要生化成分的因子分析結(jié)果

Table 4 Factor analysis of the main biochemical compositions of jasmine-scented tea by total extracting method

變量Variable因子FactorF1F2F3F4F5共同性ExtractionX'10.7490.948X'20.9130.987X'30.7060.829X'40.8250.989X'50.9040.964X'6-0.9250.866X'7-0.8880.907X'8-0.5680.7300.955X'9-0.8000.5070.928X'100.9220.931X'110.6340.546X'120.8260.948X'130.8150.739X'14-0.9350.968X'150.8950.965X'16-0.7060.5510.982X'17-0.9320.952X'18-0.9480.959X'190.6490.6150.950X'200.9290.993X'210.8620.988X'220.5510.7490.985X'230.8980.991X'240.9400.993X'250.9250.993X'260.7220.6510.985X'270.8320.992X'280.9520.993X'290.9220.981X'300.9470.935X'310.8820.957X'320.8180.936特征值Eigenvalue12.0875.8744.6974.3233.053解釋變異量Variance/%37.77118.35714.67913.5099.541累積解釋變異量Cumulativevariance/%37.77156.12970.80884.31893.858

表5為茉莉花茶全量法的因子得分矩陣,由表5及表4寫出的5個因子得分函數(shù)及1個綜合評定函數(shù)見表6。

通過這些函數(shù)即可對茉莉花茶品質(zhì)進行評定。將32個變量標準化后的值分別代入5個因子得分函數(shù),可以計算出每個因子的得分,將5個因子的得分代入綜合評定函數(shù),可以得到該茉莉花茶茶樣的綜合得分,可根據(jù)因子得分及綜合得分對待比較的茶樣間進行排名。

表5 茉莉花茶全量主要生化成分的因子得分系數(shù)矩陣

Table 5 Factor scores of the main biochemical components of jasmine-scented tea by total extracting method

變量Variable因子FactorF1F2F3F4F5X'10.0490.019-0.0310.1700.067X'2-0.0080.212-0.0010.0730.084X'3-0.0300.0720.0060.1550.129X'4-0.036-0.0190.2100.0530.000X'50.0720.0010.0100.0000.046X'6-0.112-0.0160.104-0.001-0.047X'7-0.079-0.0140.0220.041-0.061X'8-0.0560.0450.024-0.0820.296X'9-0.0850.0580.039-0.0610.220X'100.011-0.003-0.0340.265-0.178X'11-0.0380.0610.0260.191-0.089X'120.0410.038-0.059-0.0160.310X'130.0130.080-0.017-0.0330.339X'14-0.0940.0110.047-0.0070.058X'150.0180.043-0.0460.2180.044X'16-0.0600.0320.0090.1150.067X'17-0.100-0.0220.068-0.0740.022X'18-0.0920.0020.035-0.012-0.004X'190.0750.009-0.0450.1450.043X'20-0.0260.1900.0090.0150.057X'210.0280.232-0.1880.0590.082X'220.0330.149-0.025-0.0020.072X'230.0010.196-0.0070.0480.072X'24-0.0210.198-0.0100.0200.060X'25-0.081-0.0210.281-0.047-0.031X'260.014-0.0150.133-0.031-0.002X'27-0.030-0.0220.2090.022-0.014X'28-0.081-0.0490.296-0.050-0.043X'290.0780.006-0.008-0.0350.050X'300.1080.016-0.087-0.0250.060X'310.0670.0070.014-0.0410.047X'320.079-0.1210.016-0.016-0.034

表6 茉莉花茶全量浸提法品質(zhì)評定函數(shù)

Table 6 Quality evaluation functions of jasmine-scented tea by total extracting method

因子Factor函數(shù)FunctionGA和簡單兒茶素(—)GA&simplecatechins(—)F1=0.049ZX'1-0.008ZX'2-0.03ZX'3+ …+0.067ZX'31+0.079ZX'32氨基酸AminoacidF2=0.019ZX'1+0.212ZX'2+0.072ZX'3+ …+0.007ZX'31-0.121ZX'32咖啡堿CaffeineF3=-0.031ZX'1-0.001ZX'2+0.006ZX'3+ …+0.014ZX'31+0.016ZX'32酯型兒茶素和茶多酚Esterifiedcatechins&polyphenolsF4=0.17ZX'1+0.073ZX'2+0.155ZX'3+ …-0.041ZX'31-0.016ZX'32ECG,CG,C,ECF5=0.067ZX'1+0.084ZX'2+0.129ZX'3+ …+0.047ZX'31-0.034ZX'32綜合評定ComprehensiveevaluationF=(37.771F1+18.357F2+14.679F3+ 13.509F4+9.541F5)/93.858

GA：沒食子酸；ECG:表兒茶素沒食子酸酯;CG：兒茶素沒食子酸酯;C:兒茶素:EC:表兒茶素;ZX′:標準化變量值.

GA：Gallic acid； ECG: Epicatechin gallate; CG： Catechin gallate; C: Catechin; EC: Epicatechin; ZX′: Standardized variable values.

2.2 茉莉花茶價格區(qū)間的判別分析

將112個價格范圍為28～3 000元/kg的茉莉花茶茶樣分成3段,具體分段方式見表7。以2.1節(jié)中因子分析所得的公共因子為自變量,對3個價格區(qū)間進行判別分析,得到每一區(qū)間價格的判別函數(shù)(表8)。由32個變量的標準化值按照因子得分函數(shù)可計算出因子得分,再將因子得分代入判別函數(shù)即可算出每個類別的判別函數(shù)值,其中哪個類別的判別函數(shù)最大說明該茶樣屬于哪一類別。

從審評沖泡法和全量浸提法2個角度出發(fā),分別對這3個價格區(qū)間進行判別。各樣本在2個典則判別函數(shù)上得分值的空間散點圖(圖1)顯示,不同價格區(qū)間的茉莉花茶在空間分布上有顯著區(qū)別,雖然區(qū)間2和區(qū)間1、3均存在少數(shù)重疊,但這2種方法的3個價格區(qū)間的中心點都能顯著區(qū)分。

表7 價格區(qū)間的劃分

“[ ”或“]”表示包含邊緣值;“)”表示不包含邊緣值。

“[ ” or “]” indicates that the boundary value is included; “)” indicates that the boundary value isn’t included.

表8 價格區(qū)間的Bayes判別函數(shù)

A:審評沖泡法;B:全量浸提法。A: Brewing method; B: Extracting method.圖1 不同價格區(qū)間的茉莉花茶依據(jù)前2個典則判別函數(shù)的空間散點圖Fig.1 Group scatter plot of different price ranges of jasmine-scented tea on canonical discrimination functions

審評沖泡法所得判別函數(shù)的判別結(jié)果見表9。在回代判別中,價格/(元/kg)在[28,200)區(qū)間的正判率為93.8%,32個樣品中30個判別正確,2個被誤判為第2類;價格在[200,600)區(qū)間的正判率為73%,37個樣品中27個判別正確,5個被判別為第1類,5個被判別為第3類;價格在[600,3 000]區(qū)間的正判率為90.7%,43個樣品中39個判別正確,4個被判別為第2類?；卮傉新蕿?5.7%。在交叉驗證判別中,價格在[28,200)區(qū)間的正判率為93.8%,同樣32個樣品中2個被誤判為第2類;價格在[200,600)區(qū)間的正判率為70.3%,37個樣品中6個被誤判為第1類,5個被誤判為第3類;價格在[600,3 000]區(qū)間的正判率為90.7%,43個樣品中4個被誤判為第2類。交叉驗證總正判率為84.8%。說明采用審評沖泡法測得茶湯中的化學(xué)成分通過因子分析提取5個公共因子,利用這5個公共因子對茉莉花茶3個價格區(qū)間進行判別是可行的,得到的判別函數(shù)穩(wěn)定性良好,正確判斷力較高。

表9 茉莉花茶審評沖泡法價格判別結(jié)果

Table 9 Classification results of price range of jasmine-scented tea by brewing method for sensory analysis

驗證類型Testtype項目Item價格類別Type價格類別Type123總計Total原始回代Backsubstitution個數(shù)Number130203225275373043943百分率Percentage/%193.86.20100213.57313.5100309.390.7100交叉驗證Cross-validation個數(shù)Number130203226265373043943百分率Percentage/%193.86.20100216.270.313.5100309.390.7100

與審評沖泡法類似,表10為全量浸提法的價格判別結(jié)果.在回代判別中,價格在[28,200)區(qū)間的正判率為90.6%,32個樣品中29個樣品判別正確,3個被誤判為第2類;價格在[200,600)區(qū)間的正判率為78.4%,37個樣品中29個樣品判別正確,5個被誤判為第1類,3個被誤判為第3類;價格在[600,3 000]區(qū)間的正判率為83.7%,43個樣品中36個判別正確,7個被誤判為第2類?；卮傉新蕿?3.9%。在交叉驗證判別中,價格在[28,200)區(qū)間的正判率為90.6%,與回代判別相同;價格在[200,600)區(qū)間的正判率為73%,37個樣品中6個被誤判為第1類,4個被誤判為第3類;價格在[600,3 000]區(qū)間的正判率為81.4%,43個樣品中8個被誤判為第2類。交叉驗證總正判率為81.3%。說明采用全量浸提法測得的化學(xué)成分含量通過因子分析提取出5個公共因子,利用公共因子建立判別函數(shù)對茉莉花茶3個價格區(qū)間進行判別也是可行的,雖然其回代判別及交叉驗證判別的總正判率均低于審評沖泡法,但均達到了80%以上,說明所得判別函數(shù)穩(wěn)定性良好,正確判斷力也較高。

表10 茉莉花茶全量浸提法價格判別結(jié)果

Table 10 Classification results of price range of jasmine-scented tea by total extracting method

驗證類型Testtype項目Item價格類別Type價格類別Type123總計Total原始回代Backsubstitution個數(shù)Number129303225293373073643百分率Percentage/%190.69.40100213.578.48.11003016.383.7100交叉驗證Cross-validation個數(shù)Number129303226274373083543百分率Percentage/%190.69.40100216.27310.81003018.681.4100

3 討論與結(jié)論

本文以價格從28元/kg到3 000元/kg的112個茉莉花茶茶樣為實驗材料,分別從審評沖泡法和全量浸提法2個角度對茉莉花茶品質(zhì)進行因子分析并對價格區(qū)間進行判別分析,結(jié)果如下:

對于審評沖泡法,因子分析得出綜合32項理化指標91.825%信息量的5個公共因子,分別為兒茶素和GA(—)因子、茶多酚(—)因子、氨基酸因子、咖啡堿因子、CG因子,得到5個因子得分函數(shù)和1個綜合評定函數(shù);對3個價格區(qū)間的判別分析中,回代和交叉驗證的總正判率分別為85.7%和84.8%,并得到3條Bayes判別函數(shù)。

對于全量浸提法,因子分析得出綜合32項理化指標93.858%信息量的5個公共因子,分別為GA和簡單兒茶素(—)因子、氨基酸因子、咖啡堿因子、酯型兒茶素和茶多酚因子、ECG-CG-C-EC因子,同樣得到5個因子得分函數(shù)和1個綜合評定函數(shù);對3個價格區(qū)間的判別分析中,回代和交叉驗證的總正判率分別為83.9%和81.3%,并得到3條Bayes判別函數(shù)。

通過因子分析,我們將32項指標壓縮成了包含90%以上信息量的5個公共因子,并由公共因子進一步得到品質(zhì)的綜合評定函數(shù)和價格區(qū)間的判別函數(shù),從而可對茉莉花茶品質(zhì)及價格進行量化評定。由32項指標的標準化值可計算出公共因子得分,由因子得分又可計算出綜合得分及判別函數(shù)得分,最大函數(shù)值是由哪條判別函數(shù)計算得出的,就可知茶樣屬于哪個價格區(qū)間。

對于公共因子的解釋命名,本文中采用因子反映的主要物質(zhì)及與主要物質(zhì)的正負關(guān)系的方式進行命名,意在以這些物質(zhì)表征的滋味感受進行命名,如鮮爽度因子、苦澀度因子、濃強度因子等,這樣會比較直觀,直接映射在感官品質(zhì)上,但由于這些物質(zhì)與茉莉花茶滋味感覺上的具體關(guān)系仍尚未明確,雖然前人關(guān)于此方面也有一些結(jié)果呈現(xiàn),但由于其關(guān)系都有成立的范圍和條件,不可一概而論,因此暫時還是以物質(zhì)直接命名更加合理些.關(guān)于物質(zhì)與滋味感覺的關(guān)系探索仍是一個重要的研究方向,有待更加全面系統(tǒng)的研究。另外,測定的理化成分還有待進一步擴充,這對于更加全面地對品質(zhì)進行評價起到重要作用。

在判別分析中,我們將價格范圍從28元/kg到3 000元/kg劃分成了3個區(qū)間,判別效果較好,今后對價格區(qū)間更加細分的判別甚至進行價格的定量分析判別方面有待更進一步的研究,這對于指導(dǎo)市場合理定價、規(guī)范市場秩序、保護消費者及商家權(quán)益都有重要意義。

通過對審評沖泡法和全量浸提法2種茶湯獲取方法的對比,我們發(fā)現(xiàn)審評沖泡法對于價格判別的回代和交叉驗證的總正判率均高于全量浸提法,由此說明審評沖泡法對于研究茶葉品質(zhì)、構(gòu)建茶葉品質(zhì)判定模型具有可行性,且該方法簡單易行,并與日常茶葉品質(zhì)評定方法一致,關(guān)系更為直接。

[1] 孫威江,董青華,周衛(wèi)龍,等.烏龍茶品質(zhì)評定與產(chǎn)品判別研究.茶葉科學(xué),2011,31(4):305-312. Sun W J, Dong Q H, Zhou W L,etal. Study on the quality evaluation and product discriminate of Oolong tea.JournalofTeaScience, 2011,31(4):305-312. (in Chinese with English abstract)

[2] 夏怡凡.SPSS統(tǒng)計分析精要與實例詳解.北京:電子工業(yè)出版社,2010:218,243-244. Xia Y F.SPSSStatisticalAnalysisEssentialsandDetailedExamples. Beijing: Electronic Industry Press, 2010:218,243-244. (in Chinese)

[3] 楊紅,汪志君.茶湯冷后渾中生化成分含量的因子分析.食品研究與開發(fā),2006,27(8):72-75. Yang H, Wang Z J. Factor analysis of biochemistry ingredients content of tea cream of tea infuse.FoodResearchandDevelopment, 2006,27(8):72-75. (in Chinese with English abstract)

[4] 葉國注,袁海波,江用文,等.Bayes逐步判別法在綠茶板栗香化學(xué)識別上的應(yīng)用.茶葉科學(xué),2009,29(1):27-33. Ye G Z, Yuan H B, Jiang Y W,etal. Application of Bayes stepwise discrimination analysis on chemical recognition of green tea with chestnut-like aroma.JournalofTeaScience, 2009,29(1):27-33. (in Chinese with English abstract)

[5] 周利兵.我國六大名茶中微量元素的評價研究.湖南農(nóng)業(yè)科學(xué),2010(13):123-124,128. Zhou L B. Appraisal of trace element in six kinds of famous tea in China.HunanAgriculturalSciences, 2010(13):123-124,128. (in Chinese with English abstract)

[6] 成浩,王麗鴛,周健,等.基于化學(xué)指紋圖譜的扁形茶產(chǎn)地判別分析研究.茶葉科學(xué),2008,28(2):83-88. Cheng H, Wang L Y, Zhou J,etal. Discriminant classification of production area of flatten-shaped green tea based on multiple chemical fingerprint.JournalofTeaScience, 2008,28(2):83-88. (in Chinese with English abstract)

[7] 王麗鴛,成浩,周健,等.基于多元化學(xué)指紋圖譜的武夷巖茶身份判別研究.茶葉科學(xué),2010,30(2):83-88. Wang L Y, Cheng H, Zhou J,etal. Discriminant classification of Wuyi Yan tea based on multiple chemical fingerprint.JournalofTeaScience, 2010,30(2):83-88. (in Chinese with English abstract)

[8] Liang Y R, Lu J L, Zhang L Y,etal. Estimation of black tea quality by analysis of chemical composition and colour difference of tea infusions.FoodChemistry, 2003,80(2):283-290.

Quality evaluation and price discrimination of jasmine-scented tea.

Journal of Zhejiang University (Agric. & Life Sci.), 2015,41(5):577-585

Zhang Jun1,2, Gong Shuying1*, Tang Desong1*, Zhang Yingbin1,3, Chen Meili1,4

(1.TeaResearchInstitute,ZhejiangUniversity,Hangzhou310058,China; 2.HangzhouTeaResearchInstitute,AllChinaFederationofSupplyandMarketingCooperatives,Hangzhou310016,China; 3.TeaResearchInstitute,ChineseAcademyofAgriculturalSciences,Hangzhou310008,China; 4.ResearchInstituteofForestryScienceofLiuzhou,Liuzhou545300,Guangxi,China)

Jasmine-scented tea is a mainstream product of scented tea, which is very popular in Chinese northern tea market. Until nowadays, tea quality evaluation still mainly depends on the sensory assessment, which is potentially affected by individual preference and physical and mental conditions of tasters. It is necessary to search for objective and quantifiable methods for quality evaluation and discrimination. Factor analysis and discriminant analysis are statistical methods widely used in natural science, medicine, economics, sociology and many other fields. There have been some researches about application of these two methods in evaluating tea quality or discriminating the locality, variety and grade of tea. In this paper, factor analysis and discriminant analysis were used for quality evaluation and price discrimination of jasmine-scented tea. Compared with the sensory evaluation, extraction is an usual method to get tea liquor for tea quality research, which is considered to be a time and energy consuming method. In the present paper, brewing method for sensory analysis was introduced in the price discrimination study, aiming at simplifying the method for preparing tea liquor.

One hundred and twelve samples of jasmine-scented tea with price ranging from 28 to 3 000 Yuan RMB per kilogram were purchased from tea market or supplied by tea companies of Wuyutai, Zhangyiyuan, Yuanchanghou and Jinghua. The tea liquor was prepared by brewing and extracting methods respectively. Brewing method for sensory analysis, 3.0 g jasmine-scented tea was infused with 150 mL freshly boiled water for 5 min and then filtered while hot. The concentrations (mg/mL) of 13 main chemical components were measured. Total extracting method, the tea liquor was prepared according to the method of China national standard GB/T 8312—2002. Extracting 3.0 g jasmine-scented tea powder with 450 mL boiled water for 45 min in boiling water bath and shaking it in every 10 min. The liquor was pressure filtered while hot and diluted with distilled water to 500 mL when cooled. The contents (%) of 13 main chemical components were measured. Concentrations and contents of polyphenols, amino acids and water soluble extract were determined by the methods of China national standard GB/T 8313—2002, GB/T 8314—2002 and GB/T 8305—1987, respectively. While concentrations and contents of caffeine, gallic acid (GA) and the composition of catechins including gallocatechin (GC), epigallocatechin (EGC), catechin (C), epicatechin (EC), epigallocatechin gallate (EGCG), gallocatechin gallate (GCG), epicatechin gallate (ECG), catechin gallate (CG) were determined by high performance liquid chromatography (HPLC). Factor analysis and discriminant analysis were carried out based on the 32 variables using SPSS 16.0 for Windows.

The models for quality evaluation and price discrimination of jasmine-scented tea samples were established by factor and discriminant analysis based on the data obtained from brewing and extracting methods. For brewing and extracting methods, 5 factors with cumulative variance of 91.825% and 93.858% were obtained along with factor analysis, and 5 factor score functions and 1 comprehensive evaluation function were obtained. Three Bayes discrimination functions were obtained along with discriminant analysis. For the two methods, recognition accuracy of back substitution was 85.7% and 83.9%, respectively; meanwhile the recognition accuracy of the cross-validation was 84.8% and 81.3%, respectively. On the whole, the discrimination results are satisfactory and the brewing method for sensory analysis is better.

jasmine-scented tea; factor analysis; discriminant analysis; brewing method for sensory analysis； total extracting method

國家茶葉產(chǎn)業(yè)技術(shù)體系經(jīng)費資助(CARS-23)。

聯(lián)系方式:張俊(http://orcid.org/0000-0002-8533-8101),E-mail:zhangjunzi_444@126.com

2014-09-23;接受日期(Accepted):2014-12-03;網(wǎng)絡(luò)出版日期(Published online):2015-09-18

S 571.1

A

*通信作者(Corresponding authors):龔淑英(http://orcid.org/0000-0003-4424-0965),Tel:+86-571-88982519,E-mail:shuygong@zju.edu.cn;唐德松(http://orcid.org/0000-0002-1302-2255),Tel:+86-571-88982519,E-mail:tangds@zju.edu.cn

URL:http://www.cnki.net/kcms/detail/33.1247.s.20150918.1801.018.html