基于多目标优化的酿酒葡萄分级模型

Rank Model of Wine Grape Based on Multi-Objective Optimization

提出了在同时考虑酿酒葡萄和葡萄酒的情况下,运用多目标优化模型研究酿酒葡萄的分级方法.首先,建立了以酿酒葡萄的各个理化指标的误差平方和最小和酿酒葡萄对应的葡萄酒质量误差平方和最小为目标的多目标0-1规划模型;其次,利用线性加权法,将多目标0-1规划模型转化成单目标0-1规划模型;然后,建立基于酿酒葡萄分类结果的酿酒葡萄分级模型;最后,将该模型运用于酿酒葡萄分级问题中,将酿酒葡萄分成5级,每级个数分别为6,5,4,6,6,避免了分级方法的主观性和分级不均匀性.

Wine grape rank is conducted by means of multi-objective optimization model through taking into consideration the grape and the wine as well. First of all, multi-objective 0-1 programming model is established, whose objectives contain the sum of squared error of the wine grape and grape wine quality. Secondly, the multi-objective 0-1 programming model can be translated into single objective 0-1 programming model by using the linear weighting method. Then, the rank model of the wine grape is established based on the results of wine grape classification. Finally, The model is applied to the wine rank problem. The wine grapes are divided into five levels, with the number of each level being 6, 5, 4, 6, and 6. The result of experiment proves that the method successfully avoids subjectivity and uneven ranking that exist in current ranking approaches.