探讨氨基酸含量分析对羊肉产地溯源的可行性。对50个云南五地的黑山羊肉样品测定17种水解氨基酸含量，并进行方差分析、主成分分析、聚类分析、K最邻近法及人工神经网络判别分析。不同产地山羊肉Cys含量无差异，其它氨基酸含量存在不同程度差异，罗平山羊除Ala和Pro外其它氨基酸最高;石林山羊Ala和Met最高；丽江、昆明山羊Pro最高；昆明与龙陵山羊氨基酸含量则居中下水平，二者Glu、Phe、Lys、Arg、Pro含量存在显著差异（P<0.05）。主成分分析和聚类分析使羊肉样品分成不同类别，与产地基本一致。K最邻近法交叉验证对昆明样品的正确判别率为78.58%，对丽江和石林样品的正确判别率均为100%，整体正确判别率为93.48%，人工神经网络对三地样品判别正确率均为100%。对羊肉产地判别分类中人工神经网络优于K最邻近法。基于水解氨基酸含量分析对山羊肉产地来源的判别是有效可行的。

The feasibility of amino acids content analysis was studied for determination of the geographical origin of goat meat. The contents of 17 amino acids were detected in 50 hydrolyzed goat meat samples from five regions of Yunnan Province. By using analysis of variance (ANOVA), principal component analysis (PCA), cluster analysis (CA), K Nearest Neighbor (KNN) analysis and Artificial Neural Networks (ANN). The results showed that, except Cys, the amino acids contents were all different in the samples from different places. Especifically, all the amino acids contents besides Ala and Pro were the highest in the samples from Luoping area, while Ala and Met were the highest amino acid in those from Shilin area. And the content of Pro was the highest in those from Lijiang area and Kunming area. The Kunming and Longling samples had lower amino acids contents than the other samples , and these samples from the two sources showed significant differences (P＜0.05) in Glu, Phe, Lys, Arg and Pro contents.. PCA and CA classified the samples into different categories, which was almost consistent to the geographical origin. The KNN accuracy with cross-validation test reached 78.58% in Kunming samples, 100% in Lijiang samples 100% in Shilin samples as well as 93.48% in all samples. ANN provided the best results in classifying geographical origin, showing 100% accuracy in samples from Kunming, Lijiang and Shilin. Therefore, it is feasible to determine the geographical origin of goat meat by the amino acids content analysis.

公益性行业（农业）科研专项经费（201203046-5）