绿茶汤色品质感官评价的色差仪表征方法研究-论文.pdf

《绿茶汤色品质感官评价的色差仪表征方法研究-论文.pdf》由会员上传分享，免费在线阅读，更多相关内容在行业资料-天天文库。

1、河南秫业科学，2014，43(1)：149—153．JournalofHenanAgriculturalSciences绿茶汤色品质感官评价的色差仪表征方法研究吴瑞梅，赵杰文。，陈全胜，黄星奕。(1．江西农业大学工学院，江西南昌330045；2．江苏大学食品与生物工程学院，江苏镇江212013)摘要：为弥补感官审评法评定绿茶汤色品质的不足，研究了绿茶汤色品质感官评价的色差仪表征方法。采用色差仪获取茶汤的色度值，分别利用逐步回归法和主成分分析(PCA)法提取汤色特征变量，建立汤色特征变量与其感官审评

2、分值之间的相关模型。结果表明，利用PCA方法提取特征变量的效果好于利用逐步回归法。基于主成分分析法提取的特征变量，分别建立特征变量与感官评分之间的PLS和BP—ANN模型对绿茶汤色品质进行预测，其准确度可靠，但BP～ANN模型性能好于PLS模型。用BP—ANN模型对预测集样本进行测试时，预测集中各样本的预测值与实测值间相关系数为0．816，预测均方根误差为2．505。可见，利用色差仪结合BP—ANN方法能有效表征绿茶汤色品质。关键词：绿茶；茶汤汤色；感官评价；特征提取；BP神经网络法中图分类号：T

3、S272．7文献标志码：A文章编号：1004—3268(2014)01～0149—05QuantificationofSensoryEvaluationforGreenTeaQualityUsingaColorDifferenceMeterWURui—mei，ZHAOJie—wen，CHENQuan—sheng。，HUANGXing—yi。(1．CollegeofEngineering，JiangxiAgriculturalUniversity，Nanchang330045，China；2．Sch

4、oolofFood＆BiologicalEngineering，JiangsuUniversity，Zhenjiang212013，China)Abstract：Tomakeupforthedefectofsensoryevaluationmethodforevaluatinggreenteainfu—sioncolorquality，thequantificationmethodsforthesensoryevaluationofgreenteainfusioncolorbasedontheco

5、lordifferencemeterwerestudiedinthisstudy．Acolordifferencemeterwasusedtogetthecolorimetricvaluesofteainfusionforeachsample．Stepwiseregressionandprincipalcomponentanalysis(PCA)methodswererespectivelyusedtoextractthecolorimetricfeaturesfromthecolorimetri

6、cvaluesofteainfusion．Theresultsshowedthatthemodelbasedonthelea—turevariablesselectedbyPCAgotbetterresult．Thenpartialleastsquares(PLS)regressionandback—propagationartificialneuralnetwork(BP—ANN)methodswereperformedtobuildthecali—brationmodelsbasedonthe

7、relationshipbetweensensoryscoresevaluatedbyteatastersfortheinfusioncolorofteasamplesandthefeaturevariablesextractedbyPCA，respectively．TheresultsindicatedthepredictedresultsbyPLSandBP—ANNmodelswereconsistentwiththeevaluationresultsofthesensoryprofiling

8、panelsingreenteainfusioncolorstudy，buttheperformanceofBP—ANNmode1wasbetterthanthatofPIS．Thecorrelationcoefficientandrootmeansquareer—rorofpredictionofBP—ANNmodelwere0．816and2．505inthepredictionset，respectively．Thestudydemonstratesthatacolordif