基于近红外技术快速测定不同鲜肉中脂肪含量

发布时间：2018-09-14 20:45

【摘要】：随着畜禽肉和肉制品食用量的迅速增长,人们对肉品质量提出了更高的要求;对于肉制品,消费者最为关心是肉品质量,当前中国对肉品品质在线检测方面的研究和应用则相对较少,尚无针对肉品品质在线无损检测开发的设备。也没能真正投入到肉品的生产加工过程。研究不同肉品脂肪的近红外快速检测模型。并采用标准化学方法进行差异分析。通过近红外技术对猪肉、牛肉、羊肉进行扫描,采用国标法(索氏提取法)对鲜肉脂肪含量进行化学值的测定,以PLS(偏最小二乘法)作为建模方法,并通过不同的光谱预处理手段分别建立了猪牛羊肉的近红外光谱参数与样品的脂肪含量之间的对应关系模型。结果表明,对于猪肉来说,选择4 260~6 014cm~(-1)波段+一阶导+Norris所建的模型效果最好,其校正相关系数和预测相关系数分别为0.955 6和0.961 6;对于牛肉来说,选择5 226~7 343cm~(-1)波段+一阶导+S-G所建的模型效果最好,其校正相关系数和预测相关系数分别为0.923 5和0.942 7;对于羊肉来说,选择5 207~7 362cm~(-1)波段+一阶导+Norris所建的模型效果最好,其校正相关系数和预测相关系数分别为0.915 7和0.939 6;对于鲜肉来说,选选用波段为5 156~6 065cm~(-1)+二阶导+S-G所建模型效果最好,其校正相关系数和预测相关系数分别为0.916 3和0.919 4。以上所有模型的校正相关系数均大于0.91,模型都具有较高的精密度,符合不同肉制品在实际生产的需求,具有分析速度快、检测成本低、分辨率高、无损的优点。
[Abstract]:With the rapid increase in the consumption of livestock and poultry meat and meat products, people have put forward higher requirements for meat quality. For meat products, consumers are most concerned about meat quality. At present, the research and application of meat quality online inspection in China is relatively few, and there is no equipment developed for meat quality online nondestructive testing. Also did not really invest in the meat production and processing process. The near infrared fast detection model of fat in different meat products was studied. The difference analysis was carried out by standard chemical method. The pork, beef and mutton were scanned by near-infrared technique. The chemical value of fresh meat fat was determined by Soxhlet extraction method. PLS (partial least square method) was used as modeling method. The model of the relationship between the near infrared spectrum parameters and the fat content of pig beef and mutton was established by different spectral pretreatment methods. The results showed that, for pork, the model with the first order guide Norris in the band of 4 260 0 014 cm ~ (-1) was the best, its corrected correlation coefficient and predictive correlation coefficient were 0.955 6 and 0.961 6, respectively, while for beef, the correlation coefficient of correction and prediction were 0.955 6 and 0.961 6, respectively. The model established by selecting the first order guide S-G in the band of 5226 ~ (7 343cm ~ (-1) was the best, the calibration correlation coefficient and the predictive correlation coefficient were 0.923 5 and 0.942 7, respectively. For mutton, the model of the first order guide Norris in the band of 52077 367 cm ~ (-1) was the best. The corrected correlation coefficient and predicted correlation coefficient were 0.915 7 and 0.939 6, respectively, and for fresh meat, the model with second-order derivative S-G was the best, the calibration correlation coefficient and prediction correlation coefficient were 0.916 3 and 0.919 4, respectively. The calibration correlation coefficient of all the above models is greater than 0.91.The model has higher precision and meets the needs of different meat products in actual production. It has the advantages of fast analysis, low detection cost, high resolution and nondestructive.
【作者单位】：山西出入境检验检疫局;中北大学;
【基金】：山西省科技攻关项目(20150313015)资助
【分类号】：O657.33;TS251.7

【相似文献】