基于改进贝叶斯算法的卷烟叶组配方自动管控方法  

Automatic Control Method for Cigarette Leaf Group Formula Based on Improved Bayesian Algorithm

作　　者：梁海玲[1] 刘鸿[1] 李志华[1] 王艳伟[1] LIANG Hailing;LIU Hong;LI Zhihua;WANG Yanwei(Technology Center,China Tobacco Guangxi Industrial Co.,Ltd.,Nanning 530001,China)

机构地区：[1]广西中烟工业有限责任公司技术中心,南宁530001

出　　处：《自动化与仪表》2024年第11期30-33,38,共5页Automation & Instrumentation

摘　　要：由于卷烟原料复杂,卷烟叶组配方管控过程中无法准确分析原料之间的关系,提出基于改进贝叶斯算法的卷烟叶组配方自动管控方法。利用传感器并结合数据仓库技术获取历史数据,从中了解卷烟叶组原料的生产条件(输入特征),针对特征参数实施预处理,利用双重加权朴素贝叶斯算法构建管控模型,获取卷烟叶组配方并通过控制器执行配方,实现自动管控。结果表明,所研究方法得出的配方的焦油含量误差小,总氮含量、烟碱含量靠近最佳值2.6%、2.5%,自动控制效果好。Due to the complexity of cigarette raw materials,it is difficult to accurately analyze the relationship between raw materials in the control process of cigarette leaf group formula.Therefore,an automatic control method for cigarette leaf group formula based on improved Bayesian algorithm is proposed.By utilizing sensors and combining data warehouse technology to obtain historical data,we can understand the production conditions(input characteristics)of cigarette leaf group raw materials.Preprocessing is carried out based on feature parameters,and a control model is constructed using a double weighted naive Bayes algorithm.The formula for cigarette leaf group is obtained and executed through a controller to achieve automatic control.The results showed that the tar content error of the formula obtained by the research method is small,the total nitrogen content and nicotine content are close to the optimal values of 2.6%and 2.5%,and the automatic control effect is good.

关 键 词：改进贝叶斯算法 卷烟叶组配方 权值 特征参数 自动管控方法 

分 类 号：TP29[自动化与计算机技术—检测技术与自动化装置]