CAJ | 학술논문

利用2012年全国大学生数学建模竞赛A题所给数据,建立了偏最小二乘回归（PLS）分析模型,选取包括芳香物质在内的所有的酿酒葡萄和葡萄酒理化指标作为自变量,以葡萄酒的质量作为因变量建立模型.通过自变量和因变量提取成分的贡献率的计算,选取了可以解释自变量比率为86.28％的13个成分对,得到葡萄酒的质量与所有理化指标之间的PLS回归方程.利用标准化指标变量的回归方程,得到对葡萄酒的质量影响较大的理化指标有23个,其中13个正向指标,10个负向指标,正向指标中影响系数最大是葡萄的干物质含量（影响系数为0.361 4）,负向指标中影响系数最大的是葡萄的褐变度（影响系数为-0.378 8）.用Matlab编程对模型的精度进行了检验.检验结果表明如果不考虑酿酒工艺等技术、环境因素的影响,利用理化指标根据建立的偏最小二乘回归模型预测葡萄酒的质量具有很高的可信度.
이용2012년전국대학생수학건모경새A제소급수거,건립료편최소이승회귀（PLS）분석모형,선취포괄방향물질재내적소유적양주포도화포도주이화지표작위자변량,이포도주적질량작위인변량건립모형.통과자변량화인변량제취성분적공헌솔적계산,선취료가이해석자변량비솔위86.28％적13개성분대,득도포도주적질량여소유이화지표지간적PLS회귀방정.이용표준화지표변량적회귀방정,득도대포도주적질량영향교대적이화지표유23개,기중13개정향지표,10개부향지표,정향지표중영향계수최대시포도적간물질함량（영향계수위0.361 4）,부향지표중영향계수최대적시포도적갈변도（영향계수위-0.378 8）.용Matlab편정대모형적정도진행료검험.검험결과표명여과불고필양주공예등기술、배경인소적영향,이용이화지표근거건립적편최소이승회귀모형예측포도주적질량구유흔고적가신도.
This paper established a model of partial least squares regression analysis （PLS）, the use of the 2012 National Undergraduate Mathematical Contest in Modeling A title given data , including selected aromatic substances , including all of the physical and chemical indicators of wine grapes and wine as independent variables to wine quality as the dependent variable to model. Extracts calculated by the independent variables and the dependent variable contribution rate can be explained by the independent variables selected ratio was 86.28% of the 13 ingredients right, get the PLS regression equation with all the physical and chemical quality of the wine between indicators. The use of standardized indicators variable regression equation to obtain a greater impact on the quality of the wine has 23 physical and chemical indicators , the dry matter content of which 13 positive indicators , 10 negative indicators , positive indicators are the biggest factor affecting grapes （ impact factor of 0. 3614 ） , negative indicator is the biggest factor affecting the degree of browning grapes （ impact factor is -0. 3788 ） . Matlab programming for the accuracy of the model was tested . Test results show that if you do not consider the technical , environmental factors such as brewing process , the use of physical and chemical indicators have high credibility based on partial least squares regression model to predict the quality of the wine establishment .