关联规则挖掘方法在智能卷烟配方维护中的应用

Application of association rule mining method to intelligent maintenance of tobacco blend of cigarette

为了保持卷烟产品质量的稳定,需要经常通过卷烟配方维护工作来对卷烟配方中的单料烟进行替换和构成比例调整。烟叶的替换问题是卷烟配方维护的关键问题。实际生产中,一般通过经验+品吸进行替换,该方法主观性强,效率较低。本研究基于H烟草公司提供的配方调整数据,运用关联规则挖掘方法挖掘数据间的隐含规则,明确了某品牌卷烟3组主要配方以及单料烟之间的替换规则。结果说明该方法可以得到有效、可行的单料烟配伍规则,可以辅助卷烟配方维护工作。

In order to keep cigarette product’s quality stable, cigarette formula maintenance should be conducted where missing grades leaf tobacco need to be replaced, and tobacco blend of cigarette adjusted frequently. The aim of this paper is to bulid rules of replacement. Association rules mining method was employed to extract implicit rules from data, to identify main ingredients, and to acquire rules of replacement. The experiments were carried out on data of tobacco blend of cigarette from H Tobacco Company to obtain main ingredients for a cigarette brand, three groups of main tobacco blend of cigarette, and replacement rules between single tobaccos. Results indicated that this method could obtain effective rules for compatibility of single tobaccos and small number of extracted rules was comprehensible, which could support maintenance of tobacco blend of cigarette.