冷鲜猪肉新鲜度的三维荧光光谱无损检测方法

发布时间：2018-08-13 14:08

【摘要】：中国是世界猪肉生产和消费大国,冷鲜猪肉产品是我国人民猪肉消费的主要形式。在众多猪肉新鲜度无损检测方法中,荧光光谱技术以其特异性好、灵敏度高、快速、无损、检测成本低等优点受到国内外学者的关注。三维荧光光谱技术是近三十年来发展起来的一门新的荧光光谱技术,能够获得激发波长和发射波长同时变化时的荧光强度结果。本文以冷鲜猪肉为研究对象,首先对冷鲜猪肉原始三维荧光光谱进行特征分析,随后采用平行因子法分离出冷鲜猪肉原始三维荧光光谱中的荧光组分,并基于分离出的荧光组分构建特征激发发射矩阵,最后结合化学计量法开展了冷鲜猪肉新鲜度分类和新鲜度理化指标值(挥发性盐基氮、硫代巴比妥酸和pH)的无损检测方法研究。主要研究内容、结果和结论如下:(1)初步分析了冷鲜猪肉原始三维荧光光谱图中的荧光峰个数、位置及峰值的变化规律。结果显示:1)原始三维荧光光谱图中存在3个荧光峰,记为A、B和C,它们所在位置的激发波长(λEx)和发射波长(λEm)范围分别为:λEx=250-310 nm,λEm=300-400 nm;λEx=300-450 nm,λEm=400-550 nm;λEx=320-450 nm,λEm=590-610nm。2)瘦肉和肥肉中荧光峰A峰值的变化趋势相似:在20℃存储时,荧光峰A峰值在开始2天下降较快,2-4天下降较慢;在4℃存储时,在实验周期内变化较小,整体呈下降趋势。瘦肉和肥肉中荧光峰B峰值的变化趋势相似:在20℃存储时,荧光峰B峰值在开始2天上升较慢,2-4天上升较快;在4℃存储时,在实验周期内变化较小,整体呈上升趋势。(2)研究了冷鲜猪肉原始三维荧光光谱中荧光组分的分离方法,并结合前人的研究对产生荧光响应的荧光物质进行了辨析。利用平行因子分析算法从原始冷鲜猪肉三维荧光光谱中分离出了 5个主要荧光组分(C1、C2、C3、C4、C5),它们对应的荧光峰所在位置分别为:λEx=260-320nm,λEm=300-400nm;λEx=290-320 nm,λ mE300-400 nm;λEx=260-440nm,λEm=360-500nm;λEx=390-480nm,λEm=420-550nm;λEx=440-500nm,λEm=470-550nm。结合已有的文献报导,可知荧光组分C1-C4分别为:色氨酸、维生素B6、还原型辅酶(NADH)和核黄素,但荧光组分 C5(λEx=440-500nm,λEm=470-550nm)暂时未知。(3)比较了 SVM和PLS两种分类方法结合不同建模激发发射矩阵对冷鲜猪肉不同新鲜度定性分析结果的影响。结果表明:1)以相同的荧光组分所构建的单激发波长矩阵为输入变量,采用PLS方法建立的冷鲜猪肉新鲜度分类模型的分类效果优于SVM模型的分类效果。且以组分C5所构建的单激发波长矩阵(XEx5)为分类输入变量时,所建立的SVM和PLS冷鲜猪肉新鲜度定性检测模型的总体分类准确率均最高。2)在基于不同多激发波长矩阵的PLS冷鲜猪肉新鲜度分类模型中,以荧光组分C5所构建的多激发矩阵(XC5)为输入变量,所建立的冷鲜猪肉新鲜度分类模型的预测集总体分类准确率最高,为87.9%。(4)比较了 SMLR和PLSR两种定量分析方法结合不同建模激发发射矩阵对冷鲜猪肉新鲜度指标(TVB-N,TBA和pH)值定量分析结果的影响。结果表明:1)以相同荧光组分所构建的特征激发发射矩阵为建模矩阵,用PLSR方法所建立的冷鲜猪肉新鲜度指标值定量检测模型的效果均优于SMLR方法所建立的模型的效果。2)以荧光组分C4所构建的多激发波长矩阵(Xc4)为建模矩阵采用PLSR方法所建立模型最适合于TVB-N含量的定量分析(rp =0.85,RMSEP=6.91mg/100g)。以所有荧光组分所构建的全激发波长矩阵(Xall)为建模矩阵采用PLSR方法所建立模型最适合于TBA值和pH值的定量分析(rp分别为 0.80,0.90;RMSEP 分别为0.0316 mg/kg,0.2789)。
[Abstract]:China is a big country of pork production and consumption in the world. Cold and fresh pork products are the main form of pork consumption in China. Among many nondestructive detection methods for pork freshness, fluorescence spectroscopy has attracted the attention of scholars at home and abroad for its good specificity, high sensitivity, fast, nondestructive and low detection cost. A new fluorescence spectroscopy technique has been developed in the past 30 years to obtain fluorescence intensity results when excitation and emission wavelengths change simultaneously. In this paper, the original three-dimensional fluorescence spectra of chilled pork were analyzed firstly, and then the original three-dimensional fluorescence of chilled pork was separated by parallel factor method. Based on the separated fluorescent components, the characteristic excitation emission matrix was constructed. Finally, the classification of freshness and the non-destructive detection methods of physical and chemical indexes (volatile base nitrogen, thiobarbituric acid and pH) of freshness of chilled pork were studied by chemometrics. The results show that: 1) There are three fluorescence peaks in the original three-dimensional fluorescence spectrogram of chilled pork, which are called A, B and C. The excitation wavelength (lambda Ex) and emission wavelength (lambda Em) ranges are: lambda Ex = 250-310 nm, lambda Em = 300-400 nm, respectively. The fluorescence peak A of lean meat and fat meat showed similar changing trend: the peak A of lean meat and fat meat decreased rapidly at the beginning of 2 days and slowly at the beginning of 2-4 days; the fluorescence peak B of lean meat and fat meat changed slightly during the experiment period at 4 C, and the fluorescence peak B of lean meat and fat meat decreased as a whole. The variation trend was similar: the peak value of fluorescence B increased slowly in the first two days and rapidly in the second to fourth days at 20 C storage, and changed little during the experimental period at 4 C storage, and showed an overall upward trend. (2) The separation methods of fluorescence components in the original three-dimensional fluorescence spectra of chilled pork were studied, and the fluorescence response of chilled pork to the previous studies was also studied. Five main fluorescent components (C1, C2, C3, C4, C5) were separated from the original three-dimensional fluorescence spectra of chilled pork by parallel factor analysis, and their corresponding fluorescence peaks were located at: lambda Ex = 260-320 nm, lambda Em = 300-400 nm, lambda Ex = 290-320 nm, lambda mE300-400 nm, lambda Ex = 260-440 nm, lambda Em = 360-500 nm, lambda Ex = 390-320-400 nm, respectively. 480 nm, lambda Em = 420-550 nm, lambda Ex = 440-500 nm, lambda Em = 470-550 nm. According to the existing literature, the fluorescent components C1-C4 are tryptophan, vitamin B6, reduced coenzyme (NADH) and riboflavin, respectively, but the fluorescent components C5 (lambda Ex = 440-500 nm, lambda Em = 470-550 nm) are unknown for the time being. (3) The SVM and PLS classification methods are compared with different modeling excitation methods. The results showed that: 1) The single excitation wavelength matrix constructed by the same fluorescent component was used as input variable, and the classification effect of the PLS method was better than that of the SVM model. When the matrix (XEx5) was categorical input variables, the overall classification accuracy of the established SVM and PLS chilled pork freshness detection models was the highest in the overall classification rate. In the classification model of PLS fresh pork freshness based on the matrix of different excitation wavelengths, the multiple excitation matrix (XC5) constructed by fluorescence component C5 was used as the input variable, and the chilled pigs were established as the input variables. The accuracy of the model was the highest..2 The prediction set of meat freshness classification model had the highest overall classification accuracy of 87.9%. (4) The effects of SMLR and PLSR quantitative analysis methods combined with different modeling excitation emission matrices on the quantitative analysis results of fresh pork freshness indices (TVB-N, TBA and pH) were compared. The model established by PLSR method is better than that established by SMLR method. 2) The model based on the multi-excitation wavelength matrix (Xc4) constructed by fluorescent component C4 is the most suitable for quantitative analysis of TVB-N content (rp). The total excitation wavelength matrix (Xall) constructed by all fluorescent components was used as the modeling matrix. The PLSR method was most suitable for quantitative analysis of TBA and pH values (rp 0.80, 0.90, RMSEP 0.0316 mg/kg, 0.2789, respectively).
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：O657.3;TS251.51

【相似文献】