| The process of transforming "muscle" into "meat" of Tan sheep after slaughter includes three typical stages:pre-rigor,maximum rigor and ripeness.There are great differences in meat quality and processing suitability in different stages.The quality of mutton in rigor stage is poor and unedible.Therefore,it is of great industrial significance for the development of Tan mutton industry to achieve non-destructive,rapid identification and detection of meat quality in different stages of Tan sheep slaughter.In the detection of meat quality,near-infrared spectroscopy urgently needs to solve the problems of narrow scope of application and poor stability of the model,in order to achieve wide and effective application in the industry.In this paper,using Tan mutton as experimental material,the discriminant model of three different stages of postmortem muscle to meat process and the non-destructive prediction model of multiquality index of fresh Tan mutton were constructed by using visible-near infrared spectroscopy(370～1050 nm and 900～1700 nm).The rapid non-destructive identification of Tan mutton in the three stages of rigor,maximum rigor and ripeness,and the simultaneous and rapid nondestructive testing of seven quality indexes of Tan mutton color,pH value,shear force,hydraulic capacity,protein content,crude fat content and moisture content were realized.The main results are as follows:(1)Taking the meat of different parts of Tan sheep at different postmortem stages as experimental materials,partial least squares discrimination(PLS-DA)and support vector classification machine(S VMC)identification models of Tan sheep muscle at different postmortem stages were constructed based on visible-near infrared spectrum data.The results show that the discrimination effect of PLS-DA model is better than that of SVM-C model,the PLS-DA model of 370～1050 nm band original spectrum is the best,and the discrimination effect of PLS-DA model is the best after 900～1650 nm band spectrum and twoband fusion spectrum are processed by SNV.The discriminant accuracy of the optimal discriminant model established by single-band and dual-band fusion for different stages of Tan mutton samples is 100%.The determination coefficient Rp2 and root mean square error RMSEP of the optimal PLS-DA model(370～1050 nm)can reach 0.930 and 0.130.the discriminant effect and reliability of the model are good,and the discriminant effect is the best for the samples in the early stage of rigidity.(2)Taking the slaughtered meat of Tan sheep of different months as experimental materials,the partial least square regression(PLSR)prediction model of seven quality indexes and 9 parameters of Tan mutton with the fusion of sub-band and dual-band was constructed based on the visible-near infrared spectrum data.The results show that the prediction effect of PLSR model is better in 370～1050 nm band,such as a*,b*,cooking loss and protein content,and the highest correlation coefficient of prediction set is 0.922.In 900～1700 nm band,the modeling effect of L*,pH,shear force,crude fat and moisture content is better,and the highest correlation coefficient of prediction set RP is 0.941.After spectral data fusion,compared with the single-band model,the prediction effect of L*value is significantly improved,the Rp is increased to 0.900,and the improvement effect of other indicators PLSR model is not obvious.In this study,the correct discrimination rate of Tan mutton PLS-DA model at different postmortem stages can reach 100%,and the correlation coefficient R of PLSR prediction model for each quality index of Tan mutton is more than 0.80.It can realize non-destructive discrimination and rapid non-destructive prediction of Tan mutton quality indexes at different stages from muscle to meat after slaughter,in order to provide theoretical reference and technical support for postmortem meat quality control and characteristic product development and production of Tan sheep with different cooking and processing methods. |
PDF Full Text Request