基于多因素分析的烘丝机智能调控研究  

Research on Intelligent Control of Cut Tobacco Dryer Based on Multi Factor Analysis

作　　者：张风光[1] 张思明[1] 林敏[1] 叶明樵[1] 周萍芳 蒋鹏冲 刘西尧 温延[1] ZHANG Fengguang;ZHANG Siming;LIN Min;YE Mingqiao;ZHOU Pingfang;JIANG Pengchong;LIU Xiyao;WEN Yan(Wuhan Cigarette Factory,China Tobacco Hubei Industrial Co.,Ltd.,Wuhan 430040,China)

机构地区：[1]湖北中烟工业有限责任公司武汉卷烟厂,湖北武汉430040

出　　处：《控制工程》2024年第6期1138-1145,共8页Control Engineering of China

摘　　要：烘丝过程中影响出口烟丝水分和温度的干扰因素较多,为了保障出口烟丝水分与温度的稳定性,首先基于烘丝过程的历史数据,采用相关性分析方法对影响因子进行筛选。其次,选取出口烟丝水分和出口烟丝温度为目标值,通过对各机器学习模型的对比分析选取了能够快速建模与预测的极限学习机(extreme learning machine,ELM)作为建模方法,以通过模型求解运算,得出预测值。最后,采用模拟退火(simulated annealing,SA)算法,实时优化热风风速和排潮风门开度的设定值,实现对出口烟丝水分和温度的预测和控制。实验结果表明,极限学习机模型的预测效果良好,预测当前出口水分的均方根误差为0.015,出口温度的均方根误差为0.638,误差较小,保障了烘丝机智能调控方法的调控精度。During the tobacco drying process,there are numerous interfering factors that impact the moisture content and temperature of the exiting tobacco shreds.To ensure stability in these parameters,the first step involves analyzing historical process data using correlation analysis methods to identify the influential factors.Subsequently,the moisture content and temperature of the exiting tobacco shreds are selected as target values.Through comparative analysis of various machine learning models,the extreme learning machine(extreme learning machine,ELM)is chosen as the modeling method due to its rapid model development and prediction capabilities.It enables the derivation of predicted values through model calculation.Finally,the simulated annealing(simulated annealing,SA)algorithm is employed to optimize the real-time settings of hot air velocity and the valve opening of dehumidification,thereby enabling prediction and control of the moisture content and temperature of the exiting tobacco shreds.Experimental results demonstrate that the ELM model delivers excellent predictive performance with mean square error of 0.015 for predict current export moisture and an error of 0.638 for predicting exit temperature,both of which are relatively small.It ensures the precision of the intelligent control method for the tobacco dryer.

关 键 词：极限学习机 相关性分析 烘丝机 模型预测 实时优化 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]