摘要
小麦生理生化指标对研究其储藏品质具有重要的作用,但由于各指标间关系的复杂性,所表达的信息存在较大差别,这给小麦储藏品质分析带来很大的不便。针对该问题,该文提出了一种小麦储藏品质多指标分析模型,选取降落数值、发芽率、过氧化物酶、脂肪酸值、电导率、还原糖值、丙二醛7个生理生化指标作为分析的关键因素,通过相似性和主成分法对各指标进行分析计算,发现脂肪酸值最具代表性;基于脂肪酸值数据分布变化趋势,采用聚类分析方法对小麦分类;使用Fisher判别法对小麦数据进行训练,得到2类判别函数,其中判别函数1的贡献率达到89.7%,在该函数下,计算获得3种类别小麦的中心值为–5.699、1.316和3.945,从而为判断小麦的品质状况提供计算依据。试验计算结果表明,在18批储藏小麦中,该文判别模型对小麦的分类结果与实际参考标准分类结果的一致性达到88.9%,验证了本模型的合理性,研究结果可为小麦品质评价分类提供参考。
Physiological and biochemical indices play a significant role in the evaluation of wheat storage quality.The changes in the storage environment and time will not only cause the deterioration of wheat quality but also cause significant changes of wheat physiological and biochemical indices.However,the information expressed is quite different,which brings great inconvenience to the analysis of wheat storage quality.To solve this problem,a multi-index analysis model of wheat storage quality was proposed in this paper.The falling number,germination rate,peroxidase,fatty acid,conductivity,reducing sugar were selected.Seven physiological and biochemical indices of malondialdehyde were selected as the key factors in this paper.First,the KMO(kaiser-meyer-olkin)and Bartlett’s sphericity method were used to test the wheat index,and it was found that the KMO measure value was 0.807>0.7,SIG value was less than 0.001,which indicated that the selected index and the measured data were suitable for factor analysis.Then the correlation of wheat index was calculated by the European similarity coefficient and PCA(principal components analysis).The results showed that the distance between the falling number and fatty acid,reducing sugar value,malondialdehyde and conductivity was small,which indicated that the effect of these five indices on the quality of wheat was the same.Meanwhile,these five indices are very far away from germination rate and peroxidase,indicating that they are different in expressing the quality of wheat.In addition,the distance between the germination rate and peroxidase is very close.It shows that the two indices reflect the same quality in some degree.The sensitivity of fatty acid is the highest(0.186),which indicates that it has the greatest influence on the evaluation result,so this index can be used as the key index to evaluate wheat quality,and the sensitivity of peroxidase(0.160)is the least,and it has the least influence on the evaluation result.After a comprehensive analysis,the peroxidase was eliminated and the other six indices were retained for further calculation.Then,a systematic classification method based on nearest element and square Euclidean distance is used to cluster the wheat data.The classification results show that the wheat sample is composed of three kinds of data with obvious classification characteristics.The primary classification of wheat was obtained by analyzing the distribution of fatty acid data,and the discriminant function was obtained by using the Fisher discriminant method to train wheat data.According to the discriminant function,the center value of excellent wheat was–5.699.The center value of medium wheat was 1.316 and the center value of poor wheat was 3.945.By comparing the distance between the value of unknown wheat under this function and the center value of these three kinds of wheat,the unknown wheat classification can be identified.If a batch of wheat has the smallest distance to one center value of these three kinds of wheat,then it would be identified as this classification.The final test analysis shows:The result of the classification of wheat storage quality by the discriminant model in this paper is up to 88.9%in accordance with the classification of an actual reference standard.The analysis model in this paper is basically correct,which can not only provide technical support for the construction of quality evaluation system of stored wheat,but also guide the analysis and discrimination of other grain crops to a certain extent.
作者
蒋华伟
周同星
Jiang Huawei;Zhou Tongxing(School of Information Science and Engineering,Henan University of Technology,Zhengzhou 450001,China)
出处
《农业工程学报》
EI
CAS
CSCD
北大核心
2019年第10期291-298,共8页
Transactions of the Chinese Society of Agricultural Engineering
基金
国家自然科学基金(51677055)
河南省自然科学基金(162300410055)
河南省高校科技创新团队计划项目(16IRTSTHN026)
关键词
农产品
品质控制
模型
生理生化指标
相似性分析
主成分分析
系统聚类
Fisher判别法
agricultural products
quality control
models
physiological and biochemical indices
similarity analysis
principal component analysis
system clustering
Fisher discriminant method