基于主成分分析和聚类分析的不同品种燕麦品质评价
摘要：为了研究燕麦的品质并建立其评价体系，本文选取了10个燕麦品种测定其9项营养成分含量。分析了燕麦品种间各个营养指标的变异性，通过主成分分析法和聚类分析法对燕麦的品质做出综合性评价，并建立燕麦品质评价模型。结果表明，不同品种燕麦在水分、淀粉、β-葡聚糖、脂肪、总酚、黄酮等指标上存在显著性差异（P<0.05），除了水分和多不饱和脂肪酸，其他营养成分指标变异系数都大于10%。其中“花早2号”富含亚油酸，“GL380”的β-葡聚糖含量高达4.10%，“张莜14号”富含黄酮，其含量为4.14 mg RE·g FW-1，“花早2号”蛋白质含量最高为24.42%。对各品质指标进行主成分分析，提取的4个主成分累计方差贡献率达80.006%，反映了原指标的大部分信息。在所有样品中，主成分分析综合得分较高的品种为“026”、“H44”、“张莜14号”，表明这三个品种第一主成分即总酚和多不饱和脂肪酸含量相对较高，综合品质较好。为了量化样本间的相似性，减少偏离样本的干扰，在主成分分析法的基础上进行了聚类分析，聚类分析按照变量重要性（黄酮>脂肪>淀粉>总糖>水分>β-葡聚糖>蛋白质>多不饱和脂肪酸>总酚）将10种燕麦分为3类。本研究结果可为不同品种燕麦的品质评价提供借鉴，并为燕麦的育种及开发提供理论依据。
图片来源于图司机
Abstract: In order to study the nutritional value of oats and establish an oat quality evaluation system, 10 oat varieties were selected to determine contents of 9 nutritional components. The variability of various nutritional indexes among oat varieties was analyzed. The quality of oats was comprehensively evaluated by principal component analysis and cluster analysis, and the evaluation model of oat quality was established. The results showed that there were significant differences in moisture, starch, β-glucan, fat, total phenols and flavonoids among different oats (P<0.05). Except for water and polysaturated fatty acids, the coefficient of variation of other nutritional components was more than 10%. "Huazao No.2" was rich in linoleic acid, "GL380" contains up to 4.10% β-glucan, "Zhang No.14" was rich in flavonoids, containing 4.14 mg RE·g FW-1, and "Huazao No.2" had the highest protein content of 24.42%. Principal component analysis was conducted for each quality index, and the cumulative variance contribution rate of the four extracted principal components reached 80.006%, reflecting most of the information of the original index. Among all the samples, the varieties with high comprehensive scores in principal component analysis were "026", "H44" and "Zhang 14", indicating that the contents of the first principal components, namely total phenols and polyunsaturated fatty acids, were relatively high and the comprehensive quality was relatively good. In order to quantify the similarity between samples and reduce the interference of deviation samples, cluster analysis was conducted on the basis of principal component analysis. According to the importance of variables (flavonoids > fat > starch > total sugar > water > β-glucan > protein > polyunsaturated fatty acid > total phenol), 10 kinds of oats were divided into 3 categories. The results of this study can provide reference for the quality evaluation of different varieties of oats, and provide theoretical basis for the breeding and development of oats.
表1 不同品种燕麦脂肪酸组成
注：表中数据为3次重复试验的平均值与标准误差，不同字母表示同一列不同品种之间显著性差异水平（P<0.05）。
表2 不同品种燕麦营养品质
注：表中数据为3次重复试验的平均值与标准误差，不同字母表示同一列不同品种之间显著性差异水平（P<0.05）。
表3 燕麦10个品种9个营养指标的变异分析
图1 主成分分析特征值碎石图
Fig.1 Principal component analysis feature root gravel
图2 主成分分析载荷图（X、Y轴参考线原点为0）
Fig.2 Principal component load diagrams (X, Y-axis reference line origin is 0)
表4 燕麦品质评价因子的特征值和累计方差贡献率
表5 主成分在各品质指标上旋转后的成分矩阵
表6 成分得分系数矩阵
注：X1~X9分别代表水分、淀粉、总糖、葡聚糖、蛋白质、脂肪、总酚、黄酮、多不饱和脂肪酸9个指标。
图3 主成分分析得分图
表7 不同品种燕麦品质预测评价结果
注：F1、F2、F3、F4、F综分别代表第一、二、三、四主成分得分，以及综合性主成分得分。
图4 聚类分析谱系图
《食品工业科技》特邀主编专栏征稿
版权声明
《食品工业科技》具有以上论文在全世界范围内的复制权、发行权、翻译权、汇编权、广播权、表演权、信息网络传播权、转许可权及以上权利的邻接权，且作者已授权期刊同论文著作权保护期。如需转载，可联系《食品工业科技》编辑部010-87244116，或直接在文末撰写转载来源。
