红外光谱技术的三文鱼肉假冒鉴别

发布时间：2018-10-15 12:52

【摘要】：国内三文鱼市场鱼龙混杂,假冒问题严重,但鉴别方法有限。采用红外光谱技术结合偏最小二乘判别分析法(PLS-DA)研究了黑龙江大马哈鱼、淡水虹鳟、智利太平洋鲑三种鱼肉对挪威三文鱼的冒充问题。采用FITR光谱仪和KBr压片法采集四种肉类的原始光谱,并对原始光谱分别进行多元散射校正(MSC)、Savitzky-Golay平滑、一阶导数(first derivative)、标准正则变换(SNV)、峰面积归一化(peak area normalization)五种预处理来消除噪声等干扰因素并确定最佳预处理方法。为建立PLS-DA鉴别模型,将四种鱼肉的光谱分别赋予-3,-1,1和3四个参考分值,建模后通过预测检测集鱼肉得分来检验模型准确性。结果表明:采用峰面积归一化法时,PLS-DA检测模型的效果最好,校正集和交叉验证集的决定系数分别为0.97和0.95。RMSEC和RMSECV分别为0.37和0.52。该模型能显著区分四种鱼肉、检测集的预测分值分别聚集在各自的参考分值附近,在阈值为±1的判别条件下预测准确度为96%。同时采用马氏距离法进一步对四种鱼肉的光谱进行分析,发现相互之间差异明显,其中挪威三文鱼与其品种差别最大的淡水虹鳟距离最大,与其比较接近的智利太平洋鲑的距离最小,红外光谱信息能够反映不同鱼肉的品种、生活环境等差异。因此,采用红外光谱技术结合PLS-DA法能够准确的鉴别出其他鱼肉对挪威三文鱼的冒充问题,同时对其他肉类检测有一定借鉴意义。
[Abstract]:Domestic salmon market is mixed, counterfeit problem is serious, but the identification method is limited. Infrared spectroscopy combined with partial least squares discriminant analysis (PLS-DA) was used to study the impersonation of salmon from Heilongjiang, freshwater rainbow trout and Chilean Pacific salmon to Norwegian salmon. The original spectra of four kinds of meat were collected by FITR spectrometer and KBr compression method, and the original spectra were smoothed by multielement scattering correction (MSC), Savitzky-Golay, respectively. The first derivative (first derivative), standard canonical transformation (SNV), peak area normalization (peak area normalization) five pretreatments to eliminate noise and other interference factors and to determine the best pretreatment method. In order to establish the PLS-DA discriminant model, the spectra of four kinds of fish were assigned to four reference scores of -1 and 3, respectively, and the veracity of the model was tested by predicting the fish meat score. The results show that when the peak area normalization method is used, the PLS-DA detection model has the best effect, and the determination coefficients of the calibration set and the cross-validation set are 0.97 and 0.37 and 0.52 for 0.95.RMSEC and RMSECV, respectively. The model can distinguish four kinds of fish significantly, the prediction scores of the detection set are clustered around their respective reference points respectively, and the prediction accuracy is 96 when the threshold value is 卤1. At the same time, the spectrum of four kinds of fish was analyzed by Markov distance method, and it was found that there were obvious differences among them, among which the distance between Norwegian salmon and freshwater rainbow trout, which had the biggest difference in species, was the largest. The distance of Chilean Pacific salmon is the smallest, and the infrared spectrum information can reflect the difference of species and living environment of different fish. Therefore, the use of infrared spectroscopy combined with PLS-DA method can accurately identify other fish to Norway salmon impersonation, and at the same time for other meat detection has certain reference significance.
【作者单位】：华南农业大学工程学院教育部南方农业机械与装备关键技术重点实验室广东省食品质量安全重点实验室;仲恺农业工程学院信息科学与技术学院广东省食品安全与智能控制工程技术研究中心;
【基金】：国家自然科学基金青年项目(61501531) 广东省自然科学基金项目(2015A030313602) 广东省科技计划项目(2015A020209173) 广州市产学研协同创新重大专项(201508010013,201704020030)资助
【分类号】：O657.33;TS254.7

【相似文献】