【作者】 刘瑜；

【导师】 史清华；

【作者基本信息】 山东大学 ， 软件工程（专业学位）， 2013， 硕士

【摘要】 随着生活质量的提高,葡萄酒成为深受人们喜爱的饮品,葡萄酒质量的测量、分析、评价对制酒工艺的改进,和提高人们生活质量具有非常重要的意义。为了更一步的分析葡萄酒,我们深入探究葡萄酒品的基础数据,设计和实现了基于葡萄酒品基础数据的综合分析系统。该系统主要实现的功能如下。功能一,差异性检测。葡萄酒品质的好坏主要通过品酒师品尝打分给出。由于品酒师打分喜好的不同,给出的分值有差异。此系统通过单因素方差分析,分别对两组品酒师的分值进行组内和组间的差异进行显著性分析。数据显示组内差异显著,组间红酒不显著,白酒显著。为评价各组评价结果的可信性,我们建立了基于方差分析的评价模型,数据显示第二组的方差较小,即第二组的评价结果更可信。功能二,对酿酒葡萄进行分级。为了酿出好的葡萄酒,选出质量高的葡萄至关重要。我们首先利用置信区间法进行了评酒员异质性分析,通过相应的算法修正了评酒员打分,使得打分差异不再显著。同时假定不考虑葡萄酒生产工艺、储存、运输等对酒品质的影响,用修正后的得分值将对应葡萄分级,即优、一般、劣。其次,利用无量纲归一化方法,对葡萄的理化指标进行数据处理,然后采用了主成分分析的方法,得出了19个主成分,其累计贡献率超过86%。通过对每个主成分的主要组成分析,得出酿酒葡萄中最具代表性的指标是醇类、酯类、有机酸、醛类和酶类。最后通过多元回归分析,给出了各主成分和葡萄质量的回归方程,从而建立了利用葡萄的理化指标对酿酒葡萄质量的关系,试验样本检验显示,模型结果正确。功能三,发现葡萄与葡萄酒的理化指标之间的联系。我们首先利用相关系数分析法对酿酒葡萄和葡萄酒的理化指标进行关联性分析比较,生成相关系数曲面图,选取具有主要或显著相关的酿酒葡萄与葡萄酒的理化指标,并结合资料文献,确定其关联产生的原因。其次,对葡萄和葡萄酒的成分进行对比观察,给出了葡萄中成分在酿造过程中的迁移转化结果。通过分析得出：葡萄中的花色苷是红葡萄酒的关键性成色物质,存在于葡萄皮中；酿酒葡萄中的酚类和酮类通过化学反应生成酒中的化学物质并影响酒的物理性质,葡萄中的酯类、醇类在酿造过程中保持稳定,是酒的口感和香味的来源,干物质和可溶性固形物含量等则主要影响葡萄酒的亮度、色泽等物理指标。功能四,分析葡萄酒的质量与葡萄以及葡萄酒理化性质的联系。我们首先通过对理化指标的无量纲化处理,得到归一化的数据矩阵,然后通过对葡萄酒和葡萄所有理化指标的主成分分析,各自得出19个主成分。通过多元回归,得到了回归系数,从而建立了葡萄酒质量的回归预测评价模型,最后利用试验样本对预测评价结果进行了检验,验证了模型的正确性。最后,根据以上模型,我们利用C#作为实现语言,以visual studio2012作为平台,实现了葡萄酒分析系统。该系统主要实现我们以上的功能,其界面主要包括五个方面,即文件,功能,基础资料,工具,窗口和帮助。通过软件的设计与实现,我们可以将我们的模型加载到真实的系统中,使其具有更好的应用前景。更多还原

【Abstract】 With the improved quality of life, wine became very popular drink, wine quality measurement, analysis, evaluation for wine making process improvements, and improving the quality of life has a very important significance. In order to further analysis of wine, wine products we delve into the underlying data, based on the design and implementation of a comprehensive data base wine product analysis system. The main functions of the system are as follows.Features one is the difference detection. The quality of the wine tasters tasting scoring mainly through given. As sommelier scoring preferences are different, given the score differences. This system is the single factor analysis of variance, respectively, of two tasters scores for the group and the differences between groups for significance analysis. Data show significant differences within the group between the two groups was not significant red wine, white wine is remarkable. To evaluate the results of the evaluation of each group’s credibility, we have established an evaluation model based on analysis of variance, the data show a smaller variance of the second group, the second group of the evaluation results more credible.Features two is that wine grapes are graded. In order to brew good wine, selected high-quality grapes is essential. We first carried out using the confidence interval method wine appraisal heterogeneity analysis through corresponding algorithm fixes wine appraisal scoring, making scoring difference was no longer significant. Also assume not consider wine production process, storage, transportation and other effects on the quality of the wine, with the corrected score values will correspond grapes into class, namely excellent, in general, inferior. Secondly, the use of non-dimensional normalization method, the physical and chemical indicators of grape for data processing, and then using a principal component analysis method, obtained19principal components, the cumulative contribution rate of more than86%. Through each of the main components of the principal component analysis, the grape is the most representative indicators of alcohols, esters, acids, aldehydes and enzymes. Finally, multiple regression analysis, given the quality of the principal component regression equation and grapes in order to establish the physical and chemical indicators of the use of grapes for wine grape quality relationship, the test sample tests show that the model results are correct.Features three is to find grape and wine links between physical and chemical indicators. We first correlation analysis method to wine grapes and wine physicochemical indicators correlation analysis and comparison, generates correlation coefficient surface chart, select with primary or significantly associated with wine grapes and wine physical and chemical indicators, combined with information on the literature to determine its relevance causes. Secondly, the grape and wine composition were compared, gives the grapes ingredient in the brewing process of migration and transformation result. By analyzing the results:grape anthocyanins of red wine is the key color substance found in the skins of grapes; wine grapes in phenols and ketones generated by chemical reactions and chemical substances in wine affect the physical properties of the wine, Grape esters, alcohols remain stable in the brewing process, the taste and aroma of the wine source of dry matter and soluble solids content of wine mainly affect the brightness, color and other physical indicators.Features four is analyzing the quality of wine with grapes and wines physicochemical properties linked. We begin by physical and chemical indicators of the non-dimensional treatment to obtain a normalized data matrix, and then through all the physical and chemical indicators of wine and grapes principal component analysis, each drawn19principal components. Thereby establishing a regression wine quality evaluation model, the final prediction using experimental results of the evaluation sample was tested to verify the correctness of the model.Finally, according to the above model, we use C#as the implementation language to visual studio2012as a platform to achieve a wine analysis system. The system is mainly to achieve our above functions, the interface includes five aspects, namely, documents, functions, basic materials, tools, Window, and Help. Software design and implementation, we can load our model to the real system, it has a better application prospect.更多还原