基于决策树的三组元精馏序列结构最优合成规则识别

针对三组元精馏系统结构的优化合成，提出一种数据驱动的基于信息熵最小化的分类回归决策树（CART）模型。三组元精馏优化数据采用文献中严格模拟优化的结果，数据包含4种三组元混合物、34种进料组成以及7个候选精馏序列结构的最优化设计结果，生成数据集用来训练CART决策树。由决策树训练结果给出了三组元精馏结构最优设计决策规则，模型测试结果显示本文提出的决策树模型对数据集范围内三组元物系精馏结构决策的准确率为88.2%，同时表明了影响三组元最优精馏序列的主要影响因素及其重要性。

Identification of rules for optimal synthesis of ternary-distillation configuration based on decision tree

Aiming at the optimized synthesis of the structure of the three-component distillation system, a data-driven classification regression decision tree (CART) model based on information entropy minimization is proposed. Rigorous simulation based optimization results in the literature for ternary distillation configuration synthesis are used to construct data set, which include four ternary mixtures, 34 compositions and seven candidates for distillation configuration. A decision tree trained by using the data and a set of rules for the decision on the optimal ternary distillation configuration is identified. The model test result shows that the classification accuracy of the decision tree model is 88.2%, and the influential factors and the order of their importance are identified.