共查询到20条相似文献，搜索用时 203 毫秒
1.
基于矿物元素指纹的龙井茶产地溯源 总被引：4，自引：0，他引：4
龙井茶由于外形、工艺的一致性,无法通过外观特征来识别其产地,茶叶中元素组成因气候条件和土壤类型不同而有所差异,形成茶叶产地的元素指纹。本研究通过测定不同产区龙井茶中多种元素含量及同位素比率,借助化学计量学工具建立龙井茶产地区分模型。从西湖龙井、越州龙井和钱塘龙井3个产区采集102个春茶样本,采用微波消解-电感耦合等离子体质谱(ICP-MS)、电感耦合等离子体发射光谱(ICP-AES)、稳定同位素质谱(EA-IRMS)测定51种元素含量及18种元素同位素比率;运用正交偏最小二乘法判别分析(OPLS-DA)、逐步线性判别分析(FLDA)、决策树C5.0和神经网络(BP-ANN)4种判别方法对龙井茶进行产区判别。结果表明,4种方法所构建的模型对模型构建样本回代验证的正确判别率均在92%以上,而4种模型的交叉验证准确率以FLDA准确率最高,达到92.17%,决策树C5.0和BP-ANN模型略低,但也在84%以上,4种模型对外部样本的预测准确度均在80%以上,其中以OPLS-DA最高,达93.33%。不同产区龙井茶的矿质元素指纹结合化学计量学工具可以有效对龙井茶产地进行溯源;FLDA和OPLS-DA模型较适合龙井茶产地溯源。通过多元统计分析建立不同产区龙井茶判别模型,为龙井茶品牌的原产地保护提供了鉴定。 相似文献
2.
针对交叉敏感性传感器阵列电子鼻图谱含样品信息全、噪声杂、信号冗余等特点,该文根据盲信号处理原理,利用独立成分分析提取油菜蜜、椴树蜜、洋槐蜜电子鼻交叉信号中不同性质分离的独立信号。将各蜜源样本电子鼻信号矩阵延响应时间轴方向展开,统一了各样品间独立成分分解顺序的一致性,并保持了混合信号矩阵中蜂蜜样本的排序。根据不同独立成分数分离信号所带来的蜜源分类效果,8个独立成分为最优成分数。结合遗传算法筛选出各独立成分中代表蜜源样本间差异而无重复信息的特征响应点20个,来富集蜂蜜样本间的整体电子鼻差异信息,并发现大部分集中于信号的吸附阶段,少量出现于解析附阶段。油菜蜜、椴树蜜、洋槐蜜蜜源判别模型采用支持向量机算法建立,通过比较原始信号、电导变化最大值、主成分分析（principal component analysis, PCA）、独立成分分析（independent component analysis, ICA）、独立成分分析结合遗传算法（genetic algorithm, GA）这5种信号处理方式,发现ICA结合GA的特征信号挖掘效果最优,预测集判别率（95.0%）最高,其中油菜蜜、椴树蜜、洋槐蜜的预测集判别力分别为24/25、16/17、36/38;且与训练集判别率（96.3%）最接近,说明模型稳定性高、泛化能力强。结果表明该方法可以准确提取电子鼻信号中能代表蜂蜜蜜源差异信息的特征信号,并在保证蜜源分类效果的前提下,大幅减小电子鼻信号的数据量。该研究为去除电子鼻交叉信号中的冗余成分,并挖掘掩藏其下的差异信息提供了很好的指导意义,同时也拓宽了ICA的应用范围。 相似文献
3.
《农业研究与应用》2021,(3)
文章采用电子鼻对“有陈香”和“陈香不显”的两类六堡茶进行香气特征响应信号提取,以干茶、茶汤和茶底的电子鼻响应值作为香气识别模型建立的特征值,用PCA和LDA进行识别分类,并以累计贡献率最高的茶汤Loadings结果进行不同传感器贡献率的分析。结果表明PCA区分样品的累计贡献率干茶、茶汤、茶底分别达到99.29%、99.33%和94.66%,LDA区分样品的总贡献率干茶、茶汤、茶底分别为87.09%、97.50%和91.72%,LDA整体区分效果优于PCA;传感器Loadings分析结果表明:W1S (甲基类)、W2S (醇酮类)、W5S (氮氧化合物)及硫化物传感器W1W、W2W在区分样本有无陈香中起主要作用。利用PCA和LDA建立的两种模型分析方法都能将“有陈香”和“陈香不显”的六堡茶样本区分开,用随机选取的39个预测集样品分别进行识别模型的验证,干茶、茶汤、茶底的LDA识别模型对未知样本的识别准确率分别为84.62%、92.31%、82.05%,说明利用电子鼻建立判断六堡茶是否有陈香的识别模型是可行的。 相似文献
4.
近红外光谱联合CARS-PLS-LDA的山茶油检测 总被引：3，自引：0，他引：3
为了寻找快速判别山茶油掺假的检测方法,本研究利用近红外光谱技术对掺杂大豆没油山茶油进行掺假检测研究.试验在350～1 800 nm波段范围内采集样本的透射光谱,利用CARS方法筛选重要的波长变量,应用偏最小二乘-线性判别分析(PLS-LDA)建立山茶油掺假的判别模型,并与未经变量优选的判别模型进行比较.结果表明,近红外光谱技术联合CARS-PLS-LDA方法可以有效判别纯山茶油和掺假山茶油,校正集、预测集及独立样本组样本的判别正确率、灵敏度及特异性均为100％.CARS-PLS-LDA判别模型性能优于未经变量优选的判别模型,表明CARS方法可以有效筛选重要波长变量,能简化判别模型及提高判别模型的稳定性和判别精度.本研究可为山茶油掺假快速检测提供理论依据. 相似文献
5.
基于电子鼻传感器阵列优化的甜玉米种子活力检测 总被引：7，自引：5，他引：2
针对甜玉米种子活力传统检测方法操作繁琐、重复性差等不足,该研究利用电子鼻技术建立甜玉米种子活力快速检测方法。利用电子鼻获取不同活力甜玉米种子的气味信息,再结合主成分分析(PCA,principal component analysis)、线性判别分析(LDA,linear discriminant analysis)、载荷分析(loadings)和支持向量机(SVM,support vector machine)对气味信息进行提取分析,建立甜玉米种子活力的定性定量分析模型。结果显示:PCA和LDA分析均无法区分不同活力的甜玉米种子,而SVM的鉴别效果较好。全传感器阵列数据集SVM分类判别模型训练集和预测集正确率分别为97.10%和96.67%,建模时间为30.75 s,回归预测模型训练集和预测集决定系数R~2分别为0.993和0.913,均方差误差分别为2.23%和8.50%。经Loadings分析将10个传感器阵列优化为6个。优化后传感器阵列数据集SVM分类判别模型训练集和预测集正确率分别为98.55%和96.67%,建模时间为21.81 s,回归预测模型训练集和预测集决定系数R~2分别为0.982和0.984,均方差误差分别为3.80%和3.01%。结果表明:基于SVM的电子鼻技术可以实现对不同活力甜玉米种子的高效判别和预测,将传感器阵列优化为6个,判别和预测效果均有所提升。该研究为电子鼻技术应用于甜玉米种子活力检测提供理论依据。 相似文献
6.
基于主成分分析和判别分析的白酒品牌鉴别方法 总被引：3，自引：2，他引：1
白酒的香气物质决定了白酒产品的差异。为了实现不同白酒产品的区分鉴别,提出了基于气相色谱分析技术结合模式识别实现白酒区分的方法。采集了7种产品共70个白酒样本的气相色谱数据,定性定量分析了己酸乙酯、乳酸乙酯等10种基本香气物质的含量,并对测定的物质进行主成分分析,验证区分效果,最后利用线性判别法建立判别函数,对不同白酒进行区分。结果表明,2种分析方法均可用于区分不同白酒,主成分分析结果显示,前3个主成分累计贡献率为86.527%,能有效描述香气物质和产品之间的复杂关系;线性判别分析对所有样本均得到准确的判别,正确率为100%,对预测样本的正确判别率达93.9%,建立的判别函数能准确区分不同白酒。研究表明,利用气相色谱技术结合模式识别的方法可用于不同白酒的区分鉴别。 相似文献
7.
为了探索一种通过气味快速区别橙汁和橘汁以及在线监测橙汁加工品质的方法,应用电子鼻对不同品种的橙汁和橘汁的香气进行区分,研究酸橙汁加工过程中各工艺操作对香气成分的影响,并对不同加工类型的酸橙汁进行区分。通过对所获得的数据进行主成分分析及偏最小二乘回归分析,结果显示,不同品种的橙汁和橘汁的香气品质存在差异, 橙汁经过一系列加工工艺后香气发生了明显变化,浓缩还原汁的香气品质要逊于非浓缩橙汁（NFC橙汁）,28%、38%、48%和63.5% 4种不同可溶性固形物质量分数的还原汁在电子鼻传感器上的信号经拟合后有良好的线性关系,以55%还原汁为盲样,判别结果为57.95%,误差为5%。使用电子鼻可以很好的区分不同品种、不同加工类型的柑橘汁,还可以应用于橙汁加工过程中的品质控制。 相似文献
8.
为了研究超声成像技术在火腿肠质构分析与等级判别方面应用的可行性。通过对火腿肠蛋白质、淀粉等理化指标的测定将其分为特级、优级、普通级,并采集2个品牌3个等级的火腿肠共240份超声图像信息,在Matlab 7.0环境下提取图像角二阶矩、平均值等纹理特征值,最后利用线性判别式分析(linear discriminant analysis,LDA)和支持向量机(support vector machine,SVM)建立火腿肠的等级判别模型。结果表明:同品牌不同等级火腿肠超声图像、纹理特征值均具有较大差异,而同等级不同品牌火腿肠差异较小。建立的识别模型中:SVM优于LDA模型,当主成分为3时,SVM模型对应的校正集、预测集识别率均为100%,模型效果最佳。因此,超声成像技术可实现火腿肠内部质构的分析和等级的快速准确识别,研究结果可为超声成像技术在火腿肠内部质构分析和等级判别方面的应用提供参考。 相似文献
9.
茶叶中稳定同位素特征检测和溯源技术是政府职能部门市场监管和保护地理标志产品的重要手段,为评估不同烘干方式对茶叶中稳定同位素特征的影响,考察茶叶产地溯源的有效性及判别模型的稳定性,本试验通过元素分析仪联用同位素比质谱(EA-IRMS)测定5种烘干方式(烘箱直接杀青烘干、微波杀青扁形机烘干、微波杀青烘箱烘干、扁形机杀青直接烘干和扁形机杀青摊晾烘干)下西湖龙井茶(龙井群体种)中4种稳定同位素比值(δ~(13)C、δ~(15)N、δ~2H和δ~(18)O),采用单因素方差分析(one-way ANOVA)和椭圆置信区间(EJCR)测试探讨不同烘干方式下西湖龙井茶中单个和多个稳定同位素比率的差异性。单因素方差分析结果表明,不同的烘干方式可能导致茶叶中单个稳定同位素比率出现较大变化,但多因素椭圆置信区间测试表明5种烘干方式西湖龙井茶间并不存在显著性差异。基于此,建立西湖龙井茶与山东、重庆产区茶叶δ~(13)C、δ~(15)N、δ~2H和δ~(18)O的线性判别分析(LDA)模型,2 000次随机循环,3个产区茶叶判别准确度均高达90.0%以上,充分验证了溯源的有效性及判别模型的稳定性。本研究结果为茶叶原产地保护提供了理论基础与应用可行性。 相似文献
10.
依据划分苹果等级的气味指标,由气味传感器阵列和DSP构成电子鼻系统采集苹果气味信号进行处理,建立苹果气味识别模型。研究气味传感器阵列的组成形式以及其采集到的数据,设计了传感器与DSP的接口数据采集电路以及视频显示接口电路。 相似文献
11.
Young H Rossiter K Wang M Miller M 《Journal of agricultural and food chemistry》1999,47(12):5173-5177
Changes in aroma of apple harvested at four different maturities were measured at harvest and after short-term storage using electronic aroma sensors ("electronic nose") and classical headspace/gas chromatography methods. Stored fruits were also evaluated by a trained sensory panel. Compared with headspace/gas chromatography, the electronic nose was found to be more sensitive ( approximately 40 times) in terms of sample size. The sampling procedure for the electronic nose was much less complex. Using discriminant function analysis, both methods classified the apples tested into groups according to harvest date. After storage, the groupings were more diffuse. Results from sensory testing showed partial separation along the first linear discriminant but did not classify the apple into distinct groups. Important differences between treatments were found for "overall flavor", "acid flavor" intensity, "crispness", "cider/fermented aroma", "vegetative aroma", and "canned pear aroma". 相似文献
12.
【目的】利用电子鼻和分光测色仪建立一套快速检测茶树叶片氮含量的无损伤检测方法。【方法】供试样品为茶树顶芽向下第3~4片无损伤叶片。在预实验中优化了气体收集瓶体积、顶空预热温度和顶空时间等参数。采用电子鼻自带Winmuster软件将经过优化后的传感器响应特征值进行主成分分析(principal component analysis,PCA)、线性判别法分析(linear discriminant analysis,LDA)和负荷加载分析(loadings analysis,LA),筛选出灵敏性最好的传感器。同时用分光色差仪对茶树叶片色度值进行测定。样品的测量部位是叶肉区,每组20次重复。色度值主要包括L (表示黑白或者亮暗)、a (表示红绿)、b (表示黄蓝)值。采用Origin 8.0软件对测色仪L、a、b值分别进行一元线性回归分析。利用SPSS 16.0软件采用LSD法进行单因素方差分析(one-way Anova),并进行t检验。对分光测色仪中色差指标进行筛选,以获得相关系数最高的参数。采用凯氏定氮法测定茶叶总氮含量。正式试验第二步是以不同氮含量下的电子鼻和分光测色检测数据为基础,分别建立气味、颜色、气味结合颜色的3种氮含量预测模型,并进行比较分析。【结果】通过预备试验,建立了气体收集器体积为50 mL、顶空预热温度为30℃、顶空时间为30 min的电子鼻检测体系。正式试验第一步确定了以对氮氧化合物灵敏(S2),对甲烷灵敏(S6),对无机硫化物灵敏(S7),对醇类、醛类、酮类物质灵敏(S8),对有机硫化物灵敏(S9)的传感器为主要传感器。根据L、a、b表色系统,b值与叶片缺氮程度呈线性相关。正式试验第二步利用气味、颜色、气味结合颜色建立的3个氮含量预测模型都具有可行性,其中气味结合颜色建立的预测模型准确率最高,达到90%。【结论】用气味结合颜色的预测模型预测茶树叶片氮含量准确度较高,可在实际工作中进行运用。 相似文献
13.
Cheng Y Huynh-Ba T Blank I Robert F 《Journal of agricultural and food chemistry》2008,56(6):2160-2169
The temporal change in the headspace composition of an aroma model mimicking Longjing green tea aroma was studied in the presence of nonvolatile Longjing green tea constituents. Upon storage at 50 degrees C, the formation of 2-butyl-2-octenal was found, which increased with time. This enal was generated by crotonization of hexanal as demonstrated in model experiments. The formation of 2-butyl-2-octenal was also detected in Longjing tea infusions and Longjing tea leaves upon storage at 50 degrees C. The presence of nonvolatiles induced a strong decrease in aroma release. These effects were mainly due to catechins, major constituents of green tea infusion. Free amino acids, that is, glycine, contributed only to significantly decrease alpha,beta-unsaturated carbonyl aroma compounds, that is, 1-octen-3-one and geranial. Model reaction containing a mixture of 1-octen-3-one and glycine indicated on the basis of NMR and MS data the formation of the tentatively identified N-1-(3-oxo-octyl)glycine resulting from a 1,4-addition. The perceived aroma of green tea infusion is very likely to be affected by the formation of new aroma compounds and the changes in aroma release affected by interactions with tea nonvolatile components. This deserves further investigations on the sensory level. 相似文献
14.
Jonsdottir R Olafsdottir G Martinsdottir E Stefansson G 《Journal of agricultural and food chemistry》2004,52(20):6250-6256
Characterization of the flavors of ripened roe products is of importance to establish a basis for a standardized product. Flavor profiles of commercially processed ripened roe from Iceland and Norway were studied by sensory analysis, gas chromatography-olfactometry (GC-O), gas chromatography-mass spectrometry (GC-MS), and an electronic nose to characterize the headspace of ripened roe. Sensory analysis showed that ripened roe odor and flavor in combination with caviar flavor and whey/caramel-like odor give the overall positive effect of the complex characteristic roe flavor. Analysis of volatiles by GC-MS and electronic nose confirmed the presence of aroma compounds contributing to the typical ripening and spoilage flavors detected by the sensory analysis. Methional, 1-octen-3-ol, and 2,6-nonadienal were the most important compounds contributing to ripened roe odor. Spoilage flavors were partly contributed by 3-methyl-1-butanol and 3-methylbutanal, which can be measured by the electronic nose and are suggested as quality indicators for objectively assessing the ripening of roe. Principal component analysis of the overall data showed that GC-O correlated well with sensory evaluation and the electronic nose measurements. 相似文献
15.
基于电子鼻的鱼露香气品质识别 总被引：2，自引：1，他引：1
为了识别鱼露的品质,并为缩短发酵周期的工艺优选提供理论依据.利用电子鼻对7种鱼露样品的挥发性气味进行了分析,并与项空-气质联用( GC-MS)和感官分析结果进行比较.结果表明:鱼露香气成分复杂,工艺改良对气味影响很大,电子鼻18个金属传感器能很好地将不同样品的气味进行识别.以传统发酵原汁鱼露为标样,电子鼻分析结果表明,加曲改良工艺的4号样品与标样香气最为接近,相似系数达87.8％,该结果与GC-MS数据和感官分析结果一致,可为鱼露速酿工艺的优选提供参考. 相似文献
16.
为了更好地获取棉花虫害信息,该文使用电子鼻和气质联用技术对受到不同数量棉铃虫早期危害的棉花进行检测。基于气质联用技术获得了棉花挥发物的成分和含量,基于电子鼻响应曲线提取了稳定值、面积值、平均微分值、小波能量值和多项式拟合曲线参数值5种特征值,筛选出3种较优单特征:稳定值、平均微分值和面积值,之后基于多特征分别使用多层感知神经网络、径向基函数神经网络和极限学习机3种神经网络方法进行分类分析。最后采用支持向量机回归分别基于3种较优单特征及多特征对危害棉花的棉铃虫数量进行回归预测。结果表明:多特征的分类效果优于单特征,基于多特征“稳定值和平均微分值”和极限学习机分类效果最好,训练集和测试集的分类正确率均达到100%。多特征的预测能力优于单特征,基于多特征“面积值和平均微分值”的回归模型预测效果最佳,训练集回归模型的决定系数(R^2)和均方根误差(RMSE)分别为0.9940和0.0860,测试集回归模型的R^2和RMSE分别为0.9230和0.3709,电子鼻对棉花早期棉铃虫虫害具有较好的区分和预测能力,电子鼻在棉花早期棉铃虫虫害中的检测具有一定的应用潜力。 相似文献
17.
Miettinen SM Tuorila H Piironen V Vehkalahti K Hyvönen L 《Journal of agricultural and food chemistry》2002,50(15):4232-4239
The effects of emulsion structure and composition of the matrix on the release of linalool (nonpolar) and diacetyl (polar) were studied using sensory evaluation, static headspace gas chromatography, and an electronic nose. The matrices used were water, rapeseed oil, and eight oil-in-water emulsions differing in oil volume fraction (0.05/0.5), emulsifier type (sucrose stearate/modified potato starch), and homogenization pressure (100/300 bar). Fat content strongly affected the release of linalool, but it was not as critical a factor in the release of the more polar compound, diacetyl. A slight effect of the emulsifier type on the release of aromas was observed with sensory and gas chromatographic methods. The reduced droplet size, resulting from higher homogenization pressure, enhanced the release of linalool but had no effect on diacetyl. Sensory and gas chromatographic methods detected aroma changes quite similarly. The electronic nose was capable of detecting only the effect of fat on linalool. 相似文献
18.
Choi HS 《Journal of agricultural and food chemistry》2003,51(2):418-423
The volatile composition of the headspace from Citrus unshiu Marcov. forma Miyagawa-wase blossom was investigated. The volatile constituents were absorbed by a solid-phase microextraction (SPME) fiber and directly transferred to a GC-MS. Volatile compositional changes of C. unshiu blossom prepared via different drying methods (shade, microwave, and freeze-drying methods) were also determined. A total of 96 volatile constituents were confirmed in the headspace from these samples. Monoterpene hydrocarbons were prominent in the headspace volatiles of C. unshiu blossom: fresh, 84.1%; shade-dried, 60.0%; microwave-dried, 88.4%; and freeze-dried, 29.9%. p-Cymene (23.3%) was the most abundant component in the headspace of fresh C. unshiu blossom; gamma-terpinene was the most abundant in shade- and microwave-dried samples (26.8 and 31.2%, respectively) and beta-caryophyllene (10.5%) in freeze-dried sample. By using an electronic nose consisting of six metal oxide sensors, principal component analysis of the volatile compounds showed a clear aroma discrimination of the fresh and all dried blossom samples. 相似文献
19.
Martí MP Mestres M Sala C Busto O Guasch J 《Journal of agricultural and food chemistry》2003,51(27):7861-7865
The relationship between the composition and the aroma of the wine can be established by using gas chromatography with olfactometric detection (sniffing or GCO), which combines the chromatographic response with the human nose response. To evaluate the contribution of the odor compounds in wine aroma, we designed a new approach of the aroma extract dilution analysis (AEDA) that lies in the GCO analysis of serially diluted wine samples using headspace solid-phase microextraction (HS-SPME) as the extraction technique. The fiber coating used was Flex divinyl-carboxen-polydimethylsiloxane. The method developed was applied to determine the aromatic composition of a red Grenache wine from Priorat (Spain). The method allows 38 important odorants to be determined in the AEDA study, 30 of them precisely identified. These results are similar to those reported by other studies related to this variety of wine. HS-SPME is a suitable technique to obtain representative extracts of wine aroma with several advantages such as simplicity, speediness, and little sample manipulation. 相似文献
20.
Wright J Wulfert F Hort J Taylor AJ 《Journal of agricultural and food chemistry》2007,55(4):1445-1453
The release of volatile compounds from infused tea was monitored using on-line atmospheric pressure chemical ionization (APCI) mass spectrometry. Assignment of the APCI ions to particular compounds was achieved using gas chromatography of tea headspace with dual electron ionization and APCI-MS detectors. Six ions in the APCI spectrum could be assigned to individual compounds, five ions were associated with isobaric compounds (e.g., 2- and 3-methylbutanal and pentanal) or stereoisomers (e.g., heptenals or heptadienals), and a further four ions monitored were identified compounds but with some unknown impurities. Reproducibility of infusion preparation and the analytical system was good with percentage variation values generally below 5%. The analysis was used to study the effect of infusion and holding temperatures on the volatile profile of tea headspace samples, and this was found to be compound-dependent. Both the extraction of volatiles from leaf tea and the release of volatiles into the headspace play a role in creating the aroma profile that the consumer experiences. 相似文献