基于Bayes分类的原料乳中植物奶油掺假监测研究

Determination on Bayes classification used in plant cream adulterated in raw milk

对原料乳中氢化植物油和植脂末掺假进行监测。通过GC测定正常原料乳甲酯化脂肪酸,建立正常原料乳甲酯化脂肪酸数据库,用Bayes分类模型建立原料乳和氢化植物油和植脂末掺假奶的判别模型,模型所需估计的参数少、算法简单、对缺失数据不太敏感,可在SAS软件的DISCRIM过程中进行快速自动识别。建立了Bayes分类模型用于原料乳中氢化植物油和植脂末掺假奶掺假监测,得出概率判别公式为通过验证,该公式对氢化植物油和植脂末判别最低检出限分别为0.8%和0.1%。基于Bayes分类的原料乳中植物奶油掺假监测简单可行,成本低,速度快。

The aim was to study on the monitoring in plant cream adulterated into raw milk. The content of methyl fatty acid in normal raw milk determined by GC, and normal raw milk of methyl fatty acid database was established. On the bases of database the Bayes classification model was built for distinguish the raw milk, hydrogenated plant cream and the fat grain adulterated into milk. The model has the advantage of less parameters, simple algorithm and insensitive missing data. So the distinguished result could be fast and automatic in the process of SAS software. Bayes classification model was established for the quality monitoring of raw milk, the probability that a discrimination formula is R（α k│ X）= min ^R（ αk│ X） i=1,...,c. The limit of hydrogenated plant cream and the fat grain was distinguished is 0.8% and 0.1%, respectively. Bayes classification used in the monitoring of plant oil adulterated into raw milk is simple and feasible, low cost and fast.