CN104764734A - Identification method for fruit juice flavors and fragrances based on Raman spectrum and SVM algorithm - Google Patents

Identification method for fruit juice flavors and fragrances based on Raman spectrum and SVM algorithm Download PDF

Info

Publication number
CN104764734A
CN104764734A CN 201510219494 CN201510219494A CN104764734A CN 104764734 A CN104764734 A CN 104764734A CN 201510219494 CN201510219494 CN 201510219494 CN 201510219494 A CN201510219494 A CN 201510219494A CN 104764734 A CN104764734 A CN 104764734A
Authority
CN
Grant status
Application
Patent type
Prior art keywords
juice
sample
raman spectrum
raman
flavors
Prior art date
Application number
CN 201510219494
Other languages
Chinese (zh)
Inventor
童悍操
刘军
沙敏
宋超
张正勇
李大芳
Original Assignee
江苏易谱恒科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date

Links

Abstract

The invention discloses an identification method for fruit juice flavors and fragrances based on a Raman spectrum and an SVM algorithm. The method comprises the steps that firstly pretreatment is conducted on a fruit juice sample, secondly spectral data of the Raman spectrum of the fruit juice sample are collected, then analysis is conducted on the spectral data of the Raman spectrum of the fruit juice sample through the SVM algorithm, and a classification result of the flavors and fragrances of the fruit juice sample is obtained. According to the identification method for fruit juice flavors and fragrances based on the Raman spectrum and the SVM algorithm, the analysis is conducted by collecting the Raman spectral data of the fruit juice, and thereby the classification result of the fruit juice flavors and fragrances is obtained accurately, quickly and efficiently.

Description

基于拉曼光谱和SVM算法的果汁香精香料的鉴别方法 Identification of juice flavors and fragrances based on Raman spectroscopy and SVM algorithm

技术领域 FIELD

[0001] 本发明属于食品成分鉴别技术领域,特别涉及了基于拉曼光谱和SVM算法的果汁香精香料的鉴别方法。 [0001] The present invention belongs to the technical field identification food ingredients, in particular, it relates to methods for identifying a Raman spectrum juice flavor and fragrance and SVM algorithm.

背景技术 Background technique

[0002] 香料是指适合人类消费的具有香气和/或香味的物质。 [0002] refers to a spice aromas and / or aroma substances suitable for human consumption. 相对分子质量一般小于300,具有相当大的挥发性,一般不直接消费,而是配制成香精用于加香产品后间接消费。 Usually less than 300 molecular weight, having a considerable volatile, generally do not direct consumption, but indirectly formulated for consumption after fragrance perfumed product. 按用途可将香料分为日用和食用两大类。 Use can be divided into daily spices and edible two categories. 能够用于调配食品用香精的香料称为食品用香料, 包括天然香味物质、天然等同香味物质和人造香味物质三类。 Food flavor can be used in formulating perfume called food spices, flavoring substances including natural, artificial and nature-identical aroma aroma categories.

[0003] 在食品中使用食品用香料、香精的目的是使食品产生、改变或提高食品的风味。 [0003] the use of food in the food with spices, flavors purpose is to produce food, to change or enhance the flavor of food. 食品用香料一般配制成食品用香精后用于食品加香,部分也可直接用于食品加香。 General preparation of food flavoring food flavor for the food flavoring, part may be directly used in food flavoring. 食品用香精、香料在各类食品中按生产需要适量使用,GB-2760中规定巴士杀菌乳、新鲜水果和蔬菜、 原粮、大米、部分婴幼儿配方食品和饮用纯净水等食品没有加香的必要,不得添加食品用香精香料,法律、法规或国家食品安全标准另有明确规定者除外。 Food flavor, spices according to production needs appropriate to use in all kinds of foods, GB-2760 provisions bus sterilized milk, fresh fruits and vegetables, raw grain, rice, and some infant formula and other food and clean drinking water is not flavored necessary , except as otherwise expressly provided shall not add food flavor spices, laws, regulations or national food safety standards are. 该标准中还规定允许使用的食品用天然香料名单(共400种物质)和允许使用的食品用合成香料名单(共1453种物质)。 The standard also provides for allows the use of natural food flavor list (total of 400 kinds of materials) and allows the use of synthetic food flavor list (total of 1,453 kinds of matter).

[0004] 在中国,食品香精香料生广中不允许使用GB-2760食品用香料名单之外的食品香料。 [0004] In China, food flavors and fragrances Shengguang not allowed in GB-2760 Food Food spices than spices lists. 食品香料生产商、销售商和使用者都必须严格保证食品香料的质量,使用者必须保证在允许的使用量范围内使用食品香料。 Food flavoring manufacturers, vendors and users must strictly ensure the quality of food spices, the user must ensure that the use of food spices in the amount of the allowable range.

[0005] 食用香料香精质量控制一直是香料香精生产企业和食品生产企业高度关注的问题。 [0005] edible spices flavor quality control has been a problem of fragrance and flavor producers and food manufacturers a high degree of concern. 由于化学指纹图谱具有指纹特征分析、宏观推断分析等特点,故适合于分析复杂化学物质组成的稳定性,可以成为香料香精质量评价的有效手段。 Since the characteristic chemical fingerprint with fingerprint analysis, inference analysis of the macroscopic characteristics, it is suitable for analysis of complex chemical stability to the composition, it can be an effective means of evaluating the quality of spice flavors. 指纹图谱能基本反映香料香精的全貌,使其质量控制指标由原有的单一成分的测定上升为对整个内在品质的检测,实现对香精香料内在质量的综合评价和全面控制。 Fingerprints can basically reflect the whole picture spice flavors, making quality control indicators measuring some of the original single-component rose to detect the intrinsic quality of the whole, to achieve comprehensive evaluation and comprehensive control of internal quality flavors and fragrances.

[0006] 香精香料指纹图谱的仪器分析方法主要有气相色谱法、液相色谱法、紫外光谱法、 红外光谱法,各种仪器分析方法的研宄都较为成熟。 [0006] Instrumental Analysis Flavors fingerprint mainly gas chromatography, liquid chromatography, ultraviolet spectroscopy, infrared spectroscopy, a Subsidiary of various methods of analysis instruments mature. 香精香料指纹图谱一般常用的算法有改进的Nei系数法、相关系数法、夹角余弦法、距离法,其算法一般都是直接引用中药指纹图谱的算法,对香精香料指纹图谱算法进行独立研宄、比较性研宄、权重分配研宄及各算法的适应性研宄较少。 Flavors and fragrances commonly used fingerprint algorithm improved Nei coefficient, the correlation coefficient method, angle method, from the method, the algorithm is directly referenced generally Fingerprinting algorithm, flavors and fragrances of fingerprint algorithms independent study based on , comparative study based on the weight distribution and each study based on adaptive algorithm study based on less. 目前,香精香料指纹图谱技术基本上还处在研宄阶段,还没有将其直接应用于果汁生产方面的报道。 Currently, flavors and fragrances fingerprinting technology is basically still in the research stage traitor, not yet reported to be directly applied to the production of fruit juice. 因此,很有必要在现有技术的基础之上,研宄涉及一种操作便捷,能准确、快速、全面鉴定果汁用香精香料的新方法。 Therefore, it is necessary on the basis of prior art, it relates to a Subsidiary convenient to operate and can accurately and quickly, with new methods fruit juice flavors and fragrances comprehensive appraisal.

发明内容 SUMMARY

[0007] 为了解决上述背景技术提出的技术问题,本发明旨在提供基于拉曼光谱和SVM算法的果汁香精香料的鉴别方法,该方法能够准确、快速、高效地对果汁中的香精香料进行鉴别。 [0007] In order to solve the technical problems of the above-mentioned background art proposes, the present invention aims to provide authentication method juices and flavors and fragrances Raman spectrum SVM algorithm, this method can accurately, quickly and efficiently in the juice flavor and fragrance identification .

[0008] 为了实现上述技术目的,本发明的技术方案为: [0008] To achieve the above technical object, the technical solution of the present invention is:

[0009] 基于拉曼光谱和SVM算法的果汁香精香料的鉴别方法,包括以下步骤: [0009] Identification of fruit flavors and fragrances Raman spectrum and SVM algorithm, comprising the steps of:

[0010] (1)对果汁样品进行前处理,依次包括提取、富集和浓缩三个过程; [0010] (1) pre-treatment of the juice sample, comprising sequentially extracting, concentrating and three enrichment process;

[0011] (2)采集果汁样品的拉曼光谱谱图数据; [0011] (2) a Raman spectrum collected juice samples spectral data;

[0012] (3)采用SVM算法分析果汁样品的拉曼光谱谱图数据,得到果汁样品香精香料的分类结果,具体步骤为: [0012] (3) The SVM is a Raman spectrum analysis juice samples spectral data, the classification results obtained juice samples flavors and fragrances, and the specific steps:

[0013] (a)对果汁样品的拉曼光谱谱图数据进行平滑和归一化处理; [0013] (a) a Raman spectrum of spectral data juice samples smoothing and normalization process;

[0014] (b)采用PCA算法对拉曼光谱谱图数据进行特征降维,将降维后的拉曼光谱谱图数据投影到低维特征向量张成的空间中,构建训练样本集X:: [0014] (b) use PCA algorithm Raman spectroscopy spectrum data feature reduction, the dimension reduction after Raman spectrum spectral data vectors projected into a low-dimensional feature space spanned construct training sample set X: :

[0015] X= [X⑴,X(2),• • •,X(k)]GRnxk [0015] X = [X⑴, X (2), • • •, X (k)] GRnxk

[0016] 其中,k为果汁样品的种类,每类果汁样品包括n个训练样本数据, [0016] where, k is a sample of the type of juice, juice samples of each type of training data includes n,

[0017] x/1表示第i类果汁样品的第j个训练样本数据,j [0017] x / 1 represents sample i juices j-th training data, j

[0018] (c)选取核函数K,构建k(k-1) /2个SVM分类器,计算每个SVM分类器中特征向量的核函数K的值,并根据核函数K的值构成的对称矩阵计算协方差矩阵空间; [0018] (c) selecting a kernel function K, constructed k (k-1) / 2 th SVM classifier, calcd kernel function K for each feature vector in SVM classifier, based on the value and configuration of the kernel function K symmetric matrix space covariance matrix;

[0019] (d)对协方差矩阵进行Householder变换,得到对应的超平面矩阵; [0019] (d) of the covariance matrix Householder transformation, to obtain the corresponding hyperplane matrix;

[0020] (e)根据协方差矩阵和对应的超平面矩阵,计算每个特征向量的特征系数,并将特征系数对协方差矩阵进行缩放,对缩放的协方差矩阵求逆后得到模型参数《,b; [0020] (e) The hyperplane matrix and covariance matrix corresponding coefficient is calculated for each feature of the feature vector and covariance matrix coefficients of the scaling, the scaling of the covariance matrix inversion to obtain model parameters " , b;

[0021] (f)将果汁样品的测试样本数据X代入模型f(x,《) = «Tx+b中,根据f(x,《) 得到果汁样品香精香料的分类结果。 [0021] (f) the test sample data X juice samples into the model f (x, ") =« Tx + b in accordance with f (x, "juice samples to obtain classification results of flavors and fragrances).

[0022] 进一步地,步骤(1)中对果汁样品进行前处理的具体过程:取果汁样品加入提取溶剂,采用液-液萃取法进行萃取,合并萃取液,溶液过〇. 45ym滤膜后,加无水硫酸钠干燥过夜,浓缩除去提取溶剂后得到处理后的果汁样品。 [0022] Further, the step of pretreatment of the juice sample (1) in the specific procedure: Take sample of juice extraction solvent is added, by liquid - liquid extraction method, and the combined extracts were 45ym billion after the solution through the membrane, anhydrous sodium sulfate overnight, concentrated juice obtained after the extraction solvent treated samples were removed.

[0023] 进一步地,上述提取溶剂为下述一种或几种的混合物:甲醇、乙醇、乙酸乙酯、丙酮、二氯甲烷、正己烷、石油醚和乙醚。 [0023] Further, the extracting solvent is a mixture of one or more of the following: methanol, ethanol, ethyl acetate, acetone, methylene chloride, hexane, petroleum ether and diethyl ether.

[0024] 进一步地,步骤(C)中的核函数为高斯核函数[ = ,其中,p为高斯核函数的带宽参数,X1S测试样本点,X2S核函数中心。 [0024] Further, the step (C) of the kernel function is a Gaussian kernel function [=, where, p is the bandwidth parameters of the Gaussian kernel function, X1S test sample points, X2S kernel function centers.

[0025] 进一步地,采用D3便携式色散型稳频激光Raman光谱仪采集果汁样品的拉曼光谱谱图数据。 [0025] Further, Raman spectroscopy spectrum of the portable data D3 dispersive frequency stabilized laser Raman spectroscopy juice collected sample.

[0026] 进一步地,设定D3便携式色散型稳频激光Raman光谱仪的参数如下:激光光源为785nm稳频激光,激光功率为450mW,C⑶温度为制冷-85°C,光谱范围为100~3300011'系统解析度为2. 5~3.OcnT1,激光线宽为< 0. 15nm,波数矫正为+/-lcnT1,强度校正为YES,讯号质量为12000:1,重量为11kg,操作温度为(TC~50°C。 [0026] Further, the setting parameters of the laser Raman spectrometer D3 portable dispersive frequency stabilization is as follows: the laser light source is a 785nm laser frequency stabilization, a laser power of 450mW, C⑶ refrigerant temperature -85 ° C, the spectral range of 100 to 3,300,011 ' system resolution is 2. 5 ~ 3.OcnT1, the laser linewidth is <0. 15nm, wave number is correct +/- lcnT1, intensity correction is YES, the signal quality of 12,000: 1, a weight of 11kg, the operating temperature (TC ~ 50 ° C.

[0027] 进一步地,采集果汁样品的拉曼光谱谱图数据在暗室内进行,CCD冷却温度控制在-75°C~-76°C之间,扫描时间为3s,平均扫描次数为3次。 [0027] Further, Raman spectral data collected juice samples performed in a dark room, the CCD control cooling temperature between -75 ° C ~ -76 ° C, a scan time of 3s, the average number of scan 3 times.

[0028] 采用上述技术方案带来的有益效果: [0028] By adopting the technical scheme beneficial effects:

[0029]本发明通过采集果汁的拉曼光谱数据进行分析,从而得到果汁香精香料的鉴别结果,拉曼光谱提供快速、简单、可重复、且更重要的是无损伤的定性定量分析,它无需样品准备,样品可直接通过光纤探头或者通过玻璃、石英、和光纤测量。 [0029] The present invention is by Raman spectroscopy analysis of the data collected juice to give fruit juice flavor and fragrance authentication result, Raman spectroscopy offers quick, simple, reproducible, more importantly, no damage qualitative and quantitative analysis, it does not require sample preparation, the sample may be directly or via fiber optic probe through the glass, silica, and fibers. 拉曼光谱谱峰清晰尖锐,更适合定量研宄、数据库搜索、以及运用差异分析进行定性研宄。 Raman spectroscopy spectra are cleaner and sharper, more suitable for quantitative study based, database searches, and the use of qualitative study based on analysis of the differences. 本发明将香精香料指纹图谱技术直接应用于果汁生产方面,并具有检测速度快、操作简单、可重复性好、灵敏度高等优点,对果汁生产安全具有较大意义。 The flavor and fragrance of the present invention is directly applied to fingerprinting juice production, and having a fast detection, simple operation, good repeatability, high sensitivity, safety is of great significance juice.

附图说明 BRIEF DESCRIPTION

[0030] 图1是本发明的基本流程图。 [0030] FIG. 1 is a basic flow diagram of the present invention.

具体实施方式 Detailed ways

[0031] 以下将结合附图,对本发明的技术方案进行详细说明。 [0031] conjunction with the following drawings, the technical solutions of the present invention will be described in detail.

[0032] 本实施例采用3种来自不同品牌的果汁作为测试样品:样品1、样品2、样品3,并按图1所示流程进行处理。 [0032] The present embodiment uses three kinds of juice from different brands as Test Sample: Sample 1, Sample 2, Sample 3, the flow shown in FIG. 1 and press processing.

[0033] 步骤1、对样品1、样品2、样品3分别进行前处理: [0033] Step 1, for Sample 1, Sample 2, Sample 3 were pre-treatment:

[0034] 取果汁样品加入提取溶剂,采用液-液萃取法,合并萃取液,溶液过0. 45ym滤膜后,加无水硫酸钠干燥过夜,浓缩除去提取溶剂后得到供试品; [0034] Samples taken juice extraction solvent is added, by liquid - liquid extraction, the combined extracts solution was filter through 0. 45ym, anhydrous sodium sulfate overnight, and concentrated to give the test sample was removed after the extraction solvent;

[0035] 上述提取溶剂为下述一种或几种的混合物:甲醇、乙醇、乙酸乙酯、丙酮、二氯甲烷、正己烷、石油醚、乙醚; [0035] The extraction solvent is a mixture of one or more of the following: methanol, ethanol, ethyl acetate, acetone, methylene chloride, hexane, petroleum ether, diethyl ether;

[0036] 上述液-液萃取法为,取50mL果汁样品于125mL分液漏斗中,加乙醚溶液10mL,剧烈振摇提取lmin,静置分层后,从分液漏斗下部回收果汁溶液,上部倾倒出乙醚溶液,回收的果汁溶液按照上述萃取方法再重复萃取两次,合并三次萃取的乙醚溶液,经0. 45ym滤膜过滤后,加无水硫酸钠干燥过夜,过滤浓缩后加甲醇溶解并定容于5mL量瓶中得供试品溶液。 [0036] The liquid - liquid extraction method, for taking juice samples in 50mL 125mL separatory funnel, ether was added a solution of 10mL, vigorous shaking extraction Lmin, after standing layer, the juice was recovered from the bottom of the separatory funnel, pouring the upper the ether solution, a solution of the recovered juice extraction process was repeated as described above was extracted twice, and the combined ether solution was extracted three times, after membrane filtration 0. 45ym, anhydrous sodium sulfate overnight, filtered and concentrated, dissolved in methanol and set receiving in 5mL flask was the test solution.

[0037] 步骤2、分别采集样品1、样品2、样品3的拉曼光谱谱图数据: [0037] Step 2, the samples were collected 1, Sample 2, Sample 3 Raman spectrum of spectral data:

[0038] 在本实施例中,采用美国恩威广电股份有限公司生产的D3便携式色散型稳频激光Raman光谱仪进行拉曼光谱谱图数据的采集,其主要参数为:激光光源:稳频激光(785nm);激光功率:450mW;CCD温度:制冷-85°C;光谱范围:100~3300cm-1;系统解析度:2. 5~3.OcnT1;激光线宽:< 0. 15nm;波数矫正W-IcnT1;强度校正:YES;讯号质量: 12000:1 ;重量:11kg;操作温度(TC~50°C。 [0038] In the present embodiment, the United States produced Mathias Radio Co. D3 portable dispersive frequency stabilized laser Raman spectroscopy spectrum of Raman spectra collected data, its main parameters: Laser light source: a laser frequency stabilization ( of 785 nm); laser power: 450mW; CCD temperature: cooling -85 ° C; spectral range: 100 ~ 3300cm-1; system resolution:. 2 5 ~ 3.OcnT1; linewidth: <0. 15nm; corrected wavenumber W -IcnT1; intensity correction: YES; signal quality: 12,000: 1; weight: 11kg; operating temperature (TC ~ 50 ° C.

[0039] 拉曼光谱采集在暗室内进行,C⑶冷却温度控制在-75°C~-76°C之间,扫描时间为3s,平均扫描次数为3次。 [0039] Raman spectra acquisition performed in a dark room, C⑶ controlled cooling temperature between -75 ° C ~ -76 ° C, a scan time of 3s, the average number of scan 3 times.

[0040] 步骤3、采用SVM算法分析果汁样品的拉曼光谱谱图数据,得到果汁样品香精香料的分类结果: [0040] Step 3, using the SVM algorithm Raman spectroscopy spectrum data analysis of the sample of juice, juice samples to obtain classification results of flavor and fragrance:

[0041] (a)对果汁样品的拉曼光谱谱图数据进行平滑和归一化处理; [0041] (a) a Raman spectrum of spectral data juice samples smoothing and normalization process;

[0042] (b)采用PCA算法对拉曼光谱谱图数据进行特征降维,将降维后的拉曼光谱谱图数据投影到低维特征向量张成的空间中,构建训练样本集X: [0042] (b) use PCA algorithm Raman spectroscopy spectrum data feature reduction, the dimension reduction after Raman spectrum spectral data vectors projected into a low-dimensional feature space spanned construct training sample set X:

[0043] X= [X⑴,X(2),• • •,X(k)]GRnxk [0043] X = [X⑴, X (2), • • •, X (k)] GRnxk

[0044] 其中,k为果汁样品的种类,每类果汁样品包括n个训练样本数据, = .Y/("表示第i类果汁样品的第j个训练样本数据,j= 1,2,…,n ; [0044] where, k is a sample of the type of juice, juice samples of each type of training data includes n, = .Y / ( "j denotes the i-th training data juices samples, j = 1,2, ... , n;

[0045] (c)选取核函数K,构建k(k-1) /2个SVM分类器,计算每个SVM分类器中特征向量的核函数K的值,并根据核函数K的值构成的对称矩阵计算协方差矩阵空间; [0045] (c) selecting a kernel function K, constructed k (k-1) / 2 th SVM classifier, calcd kernel function K for each feature vector in SVM classifier, based on the value and configuration of the kernel function K symmetric matrix space covariance matrix;

[0046] (d)对协方差矩阵进行Householder变换,得到对应的超平面矩阵; [0046] (d) of the covariance matrix Householder transformation, to obtain the corresponding hyperplane matrix;

[0047] (e)根据协方差矩阵和对应的超平面矩阵,计算每个特征向量的特征系数,并将特征系数对协方差矩阵进行缩放,对缩放的协方差矩阵求逆后得到模型参数《,b。 [0047] (e) The hyperplane matrix and covariance matrix corresponding coefficient is calculated for each feature of the feature vector and covariance matrix coefficients of the scaling, the scaling of the covariance matrix inversion to obtain model parameters " , b.

[0048] (f)将果汁样品的测试样本数据X代入模型f(x,《) = «Tx+b中,根据f(x,《) 得到果汁样品香精香料的分类结果。 [0048] (f) the test sample data X juice samples into the model f (x, ") =« Tx + b in accordance with f (x, "juice samples to obtain classification results of flavors and fragrances).

[0049] 在本实施例中,步骤(C)中的核函数为高斯核函数尤=,其中,p为高斯核函数的带宽参数,X1S测试样本点,X2为核函数中心。 [0049] embodiment, step (C) is a Gaussian kernel function kernel function, especially in the present embodiment =, where, p is the bandwidth parameters of the Gaussian kernel function, X1S test sample points, an X2 is a function of a central core.

[0050] 以上实施例仅为说明本发明的技术思想,不能以此限定本发明的保护范围,凡是按照本发明提出的技术思想,在技术方案基础上所做的任何改动,均落入本发明保护范围之内。 [0050] The above embodiments are merely illustrative of the technical idea of ​​the present invention, in order not to limit the scope of the present invention, all made in accordance with the technical idea of ​​the present invention, any changes made on the basis of the aspect, the present invention fall within the within the scope of protection.

Claims (7)

  1. 1. 基于拉曼光谱和SVM算法的果汁香精香料的鉴别方法,其特征在于,包括以下步骤: (1) 对果汁样品进行前处理,依次包括提取、富集和浓缩三个过程; (2) 采集果汁样品的拉曼光谱谱图数据; (3) 采用SVM算法分析果汁样品的拉曼光谱谱图数据,得到果汁样品香精香料的分类结果,具体步骤为: (a) 对果汁样品的拉曼光谱谱图数据进行平滑和归一化处理; (b) 采用PCA算法对拉曼光谱谱图数据进行特征降维,将降维后的拉曼光谱谱图数据投影到低维特征向量张成的空间中,构建训练样本集X : X = [X(1),X(2),... ,X(k)] e Rnxk 其中,k为果汁样品的种类,每类果汁样品包括η个训练样本数据, X(i)= [Xl(i),x2(i),...,xn (i)],i = l,2,...,k,Xj(i)表示第i 类果汁样品的第j 个训练样本数据,j = 1,2,…,η ; (c) 选取核函数Κ,构建k (k-1) /2个SVM分类器,计算每个SVM分类器中 1. Identification of fruit flavors and fragrances Raman spectrum and SVM algorithm, characterized by comprising the steps of: (1) pre-treatment of the juice sample, comprising sequentially extracting, concentrating and enriching three processes; (2) Raman spectrum of spectral data collected sample of juice; (3) the spectral data of SVM Raman spectrum analysis of a sample of the juice, the juice samples to obtain classification results of flavor and fragrance, is the specific steps: (a) a Raman sample of juice Spectroscopic data for spectral smoothing and normalization process; (b) Raman spectrum using PCA algorithm spectral data for feature reduction, lowering the Raman spectroscopy spectrum of the dimensional data projected onto a low-dimensional feature vector spanned space, constructing training sample set X: X = [X (1), X (2), ..., X (k)] e Rnxk wherein, k is a sample of the type of juice, juice samples comprising each type of training η sample data, X (i) = [Xl (i), x2 (i), ..., xn (i)], i = l, 2, ..., k, Xj (i) denotes an i juices j-th training data samples, j = 1,2, ..., η; (c) selecting a kernel function K0, constructed k (k-1) / 2 two SVM classifiers, SVM classifier is calculated for each 征向量的核函数K的值,并根据核函数K的值构成的对称矩阵计算协方差矩阵空间; (d) 对协方差矩阵进行Householder变换,得到对应的超平面矩阵; (e) 根据协方差矩阵和对应的超平面矩阵,计算每个特征向量的特征系数,并将特征系数对协方差矩阵进行缩放,对缩放的协方差矩阵求逆后得到模型参数ω,b ; (f) 将果汁样品的测试样本数据X代入模型f (X,ω) = coTx+b中,根据f (X,ω)得到果汁样品香精香料的分类结果。 Kernel function K of the eigenvector value, and the calculation of the covariance matrix of the spatial The symmetric matrix value of the kernel function K composed; (d) of the covariance matrix Householder transformation, to give hyperplane matrix corresponding; (e) from the covariance matrix and the matrix corresponding to the hyperplane, the coefficient is calculated for each feature of the feature vector and covariance matrix coefficients of the scaling, the scaling of the covariance matrix inversion to obtain the model parameters ω, b; (f) the juice samples the test sample data X into the model f (X, ω) = coTx + b, the classification result obtained under juice samples flavors f (X, ω).
  2. 2. 根据权利要求1所述基于拉曼光谱和SVM算法的果汁香精香料的鉴别方法,其特征在于,步骤(1)中对果汁样品进行前处理的具体过程:取果汁样品加入提取溶剂,采用液-液萃取法进行萃取,合并萃取液,溶液过0. 45 μ m滤膜后,加无水硫酸钠干燥过夜,浓缩除去提取溶剂后得到处理后的果汁样品。 The identification method of claim 1 juices and flavors and fragrances Raman spectrum SVM algorithm, wherein the step of the specific process of the juice sample pretreatment (1): extracting solvent added to the juice samples taken, using for liquid extraction. the combined extracts solution was filter through 0. 45 μ m, was added over anhydrous sodium sulfate overnight, concentrated juice obtained after the extraction solvent to remove the sample after the treatment - liquid.
  3. 3. 根据权利要求2所述基于拉曼光谱和SVM算法的果汁香精香料的鉴别方法,其特征在于:所述提取溶剂为下述一种或几种的混合物:甲醇、乙醇、乙酸乙酯、丙酮、二氯甲烷、 正己烷、石油醚和乙醚。 3. The authentication method 2 juice flavor and fragrance Raman spectrum and SVM algorithm, according to claim wherein: the extraction solvent is one or a mixture of several of the following: methanol, ethanol, ethyl acetate, acetone, methylene chloride, hexane, petroleum ether and diethyl ether.
  4. 4. 根据权利要求1所述基于拉曼光谱和SVM算法的果汁香精香料的鉴别方法,其特征在于:步骤(c)中的核函数为高斯核函数I = 其中,p为高斯核函数的带宽参数, 测试样本点,X 2为核函数中心。 4. The method of identifying a Raman spectrum juice flavor and fragrance and SVM algorithm, according to claim wherein: step (c) of the kernel function is a Gaussian kernel function I = where, p is the bandwidth of the Gaussian kernel parameters, test sample points, X 2 is a function of a central core.
  5. 5. 根据权利要求1所述基于拉曼光谱和SVM算法的果汁香精香料的鉴别方法,其特征在于:采用D3便携式色散型稳频激光Raman光谱仪采集果汁样品的拉曼光谱谱图数据。 The authentication method according to a Raman spectrum fruit flavors and fragrances and SVM algorithm, as claimed in claim wherein: D3 using stabilized lasers portable dispersive Raman spectroscopy Raman spectra of the spectrometer collected data juice sample.
  6. 6. 根据权利要求5所述基于拉曼光谱和SVM算法的果汁香精香料的鉴别方法,其特征在于,设定D3便携式色散型稳频激光Raman光谱仪的参数如下:激光光源为785nm稳频激光,激光功率为450mW,C⑶温度为制冷-85 °C,光谱范围为100~3300(31^1,系统解析度为2. 5~3. OcnT1,激光线宽为< 0. 15nm,波数矫正为+/-lcnT1,强度校正为YES,讯号质量为12000:1,重量为I lkg,操作温度为0°C~50°C。 The authentication method 5 Raman spectrum juice flavor and fragrance and SVM algorithm, wherein the parameter set forth in claim D3 Laser Raman spectrometer portable dispersive frequency stabilization is as follows: the laser light source is a 785nm laser frequency stabilization, a laser power of 450mW, C⑶ refrigerant temperature -85 ° C, the spectral range of 100 to 3300 (31 ^ 1, the system resolution is 2. 5 ~ 3. OcnT1, the laser linewidth is <0. 15nm, correction of wave number + / -lcnT1, intensity correction is YES, the signal quality of 12,000: 1, by weight of I lkg, the operating temperature is 0 ° C ~ 50 ° C.
  7. 7. 根据权利要求1所述基于拉曼光谱和SVM算法的果汁香精香料的鉴别方法, 其特征在于:采集果汁样品的拉曼光谱谱图数据在暗室内进行,CCD冷却温度控制在-75°C~-76°C之间,扫描时间为3s,平均扫描次数为3次。 The authentication method according to a Raman spectrum fruit flavors and fragrances and SVM algorithm, as claimed in claim wherein: the Raman spectra collected spectral data juice sample was in a dark room, the CCD is cooled at -75 ° temperature control between C ~ -76 ° C, a scan time of 3s, the average number of scan 3 times.
CN 201510219494 2015-04-30 2015-04-30 Identification method for fruit juice flavors and fragrances based on Raman spectrum and SVM algorithm CN104764734A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 201510219494 CN104764734A (en) 2015-04-30 2015-04-30 Identification method for fruit juice flavors and fragrances based on Raman spectrum and SVM algorithm

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 201510219494 CN104764734A (en) 2015-04-30 2015-04-30 Identification method for fruit juice flavors and fragrances based on Raman spectrum and SVM algorithm

Publications (1)

Publication Number Publication Date
CN104764734A true true CN104764734A (en) 2015-07-08

Family

ID=53646698

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 201510219494 CN104764734A (en) 2015-04-30 2015-04-30 Identification method for fruit juice flavors and fragrances based on Raman spectrum and SVM algorithm

Country Status (1)

Country Link
CN (1) CN104764734A (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5261410A (en) * 1991-02-07 1993-11-16 Alfano Robert R Method for determining if a tissue is a malignant tumor tissue, a benign tumor tissue, or a normal or benign tissue using Raman spectroscopy
US6040191A (en) * 1996-06-13 2000-03-21 Grow; Ann E. Raman spectroscopic method for determining the ligand binding capacity of biologicals
CN103293141A (en) * 2013-03-25 2013-09-11 江苏省质量安全工程研究院 A liquor vintage recognition method based on a fusion technology of ion mobility spectrometry/ mass spectrometry/ Raman spectroscopy
CN104502320A (en) * 2014-12-29 2015-04-08 江南大学 Method for identifying strong flavor Baijiu by combining three-dimensional fluorescence spectrum with PCA-SVM

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5261410A (en) * 1991-02-07 1993-11-16 Alfano Robert R Method for determining if a tissue is a malignant tumor tissue, a benign tumor tissue, or a normal or benign tissue using Raman spectroscopy
US6040191A (en) * 1996-06-13 2000-03-21 Grow; Ann E. Raman spectroscopic method for determining the ligand binding capacity of biologicals
CN103293141A (en) * 2013-03-25 2013-09-11 江苏省质量安全工程研究院 A liquor vintage recognition method based on a fusion technology of ion mobility spectrometry/ mass spectrometry/ Raman spectroscopy
CN104502320A (en) * 2014-12-29 2015-04-08 江南大学 Method for identifying strong flavor Baijiu by combining three-dimensional fluorescence spectrum with PCA-SVM

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
张石清 等: "支持向量机应用于语音情感识别的研究", 《声学技术》 *
李航: "《统计学习方法》", 31 March 2012 *

Similar Documents

Publication Publication Date Title
Qin et al. Hyperspectral and multispectral imaging for evaluating food safety and quality
Wang et al. Feasibility study of quantifying and discriminating soybean oil adulteration in camellia oils by attenuated total reflectance MIR and fiber optic diffuse reflectance NIR
Karoui et al. Mid-infrared spectroscopy coupled with chemometrics: A tool for the analysis of intact food systems and the exploration of their molecular structure− Quality relationships− A review
Huang et al. Nondestructive measurement of total volatile basic nitrogen (TVB-N) in pork meat by integrating near infrared spectroscopy, computer vision and electronic nose techniques
Sliwinska et al. Food analysis using artificial senses
Shiroma et al. Application of NIR and MIR spectroscopy in quality control of potato chips
Arvanitoyannis et al. Application of quality control methods for assessing wine authenticity: Use of multivariate analysis (chemometrics)
Brescia et al. Characterisation of the geographical origin of buffalo milk and mozzarella cheese by means of analytical and spectroscopic determinations
Yu et al. A feasibility study on on-line determination of rice wine composition by Vis–NIR spectroscopy and least-squares support vector machines
Cubero-Leon et al. Review on metabolomics for food authentication
Cen et al. Measurement of soluble solids contents and pH in orange juice using chemometrics and Vis− NIRS
Danezis et al. Food authentication: Techniques, trends & emerging approaches
Nieuwoudt et al. Principal component analysis applied to Fourier transform infrared spectroscopy for the design of calibration sets for glycerol prediction models in wine and for the detection and classification of outlier samples
Downey et al. Detection of honey adulteration by addition of fructose and glucose using near infrared transflectance spectroscopy
Cozzolino Near infrared spectroscopy in natural products analysis
Cuadros-Inostroza et al. Discrimination of wine attributes by metabolome analysis
Reid et al. Differentiation of apple juice samples on the basis of heat treatment and variety using chemometric analysis of MIR and NIR data
Qin et al. Evaluation of Chinese tea by the electronic nose and gas chromatography–mass spectrometry: Correlation with sensory properties and classification according to grade level
Araghipour et al. Geographical origin classification of olive oils by PTR-MS
Jensen et al. Evaluation of quality changes in walnut kernels (Juglans regia L.) by Vis/NIR spectroscopy
Defernez et al. Use of infrared spectroscopy and chemometrics for the authentication of fruit purees
Casale et al. Study of the aging and oxidation processes of vinegar samples from different origins during storage by near-infrared spectroscopy
Boscaini et al. Gas Chromatography− Olfactometry (GC-O) and Proton Transfer Reaction− Mass Spectrometry (PTR-MS) Analysis of the Flavor Profile of Grana Padano, Parmigiano Reggiano, and Grana Trentino Cheeses
Baeten et al. Edible oils and fats authentication by Fourier transform Raman spectrometry
Falcão et al. A survey of seasonal temperatures and vineyard altitude influences on 2-methoxy-3-isobutylpyrazine, C13-norisoprenoids, and the sensory profile of Brazilian Cabernet Sauvignon wines

Legal Events

Date Code Title Description
C06 Publication
EXSB Decision made by sipo to initiate substantive examination
COR Change of bibliographic data
RJ01