基于决策树和正交试验设计的打叶复烤工艺参数优化

Parameters Optimization in Threshing and Redrying Process Based on Decision Tree and Orthogonal Experiment Design

打叶复烤过程的工艺参数设置是稳定烟叶质量的关键,为了提高烟叶质量的稳定性,本文提出了一种基于决策树与正交试验设计的参数优化方法。使用决策树算法模型挖掘与分析打叶复烤历史数据的工艺参数与出口水分指标之间的关系,得到工艺参数对指标的影响权重排序,由此设计因素水平表进行正交试验设计,根据试验结果对决策树模型进行验证与修正,并获得较优参数组合,现场验证表明修正后的预测结果与实际相比误差小于5%,所获得的参数组合充分满足现场需求。

The setting of technological parameters in threshing and redrying process is the key to stabilize the quality of tobacco leaves.In order to improve the stability of tobacco quality,a parameter optimization method based on decision tree and orthogonal experimental design was proposed.The decision tree algorithm model was used to mine and analyze the relationship between the process parameters of threshing and redrying and the export moisture index from historical data,and the influence weight of the process parameters on the indexes was obtained.The orthogonal experiment design was carried out by designing the factor level table.According to the experimental results,the decision tree model was verified and modified,and the optimal combination of parameters was obtained.The field verification showed that the modified model was feasible since the error between prediction and the actual results was less than 5%,and the obtained parameter combination fully met the field demand.