烘丝筒出口叶丝含水率预测模型研究  

Research on Moisture Content Predict Model of Tobacco Exported from Drying Cylinder

作　　者：王乐军[1] 王林枝 牛燕丽[1] WANG Lejun;WANG Linzhi;NIU Yanli(Wuhan Cigarette Factory,China Tobacco Hubei Industrial Co.,Ltd.,Wuhan 430000,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430000

出　　处：《自动化仪表》2024年第4期62-66,70,共6页Process Automation Instrumentation

摘　　要：烘丝的最佳工艺参数难以确认,且叶丝含水率预测误差较大。为了在信息技术方面辅助提升烟草成品质量,研究基于极限学习机(ELM)的烘丝筒出口叶丝含水率预测模型。选取叶丝烘丝过程中松散回潮、预混柜、润叶加料等工艺阶段环境温度、湿度、加水比例等工艺参数。通过随机森林方法,将处理后有效数据中的各烘丝工艺参数以平均精准度逐渐减少顺序进行重新排序,筛选出对烘丝筒叶丝含水率预测作用较大的烘丝工艺参数。将筛选后的烘丝工艺参数作为ELM的输入数据,获取叶丝含水率预测结果。以含水率预测平均绝对误差最小为差分进化算法的适应度函数,优化ELM的隐含层神经元数量,提升烘丝筒出口叶丝含水率预测精度。试验结果表明,该模型可实现烘丝筒出口叶丝含水率预测,且预测误差小于0.3%,预测精度高。该研究有助于提升烟草质量。The optimal process parameters for drying are difficult to confirm,and the prediction error of tabacco moisture content is large.To assist in improving the quality of finished tobacco products in terms of information technology,the extreme learning machine(ELM)-based model for predicting the moisture content of tobacco exported from drying cylinder is studied.The environmental temperature,humidity,water addition ratio and other process parameters of tobacco drying process in the loose moisture back,pre-mixing cabinet,wet leaf feeding process are selected.Through the random forest method,the processed effective data of the drying process parameters,are reordered by gradually reduce the order of the average accuracy,so that process parameters of the drying cylinder tobacco moisture content prediction of the role of the larger drying can be screened out.The filtered drying process parameters are used as input data for ELM to obtain the prediction results of tobacco moisture content.The number of neurons in the hidden layer of the ELM is optimized by taking the minimum average absolute error of moisture content prediction as the fitness function of the differential evolutionary algorithm,to improve the prediction accuracy of the moisture content of the tobacco at the exit of the drying cylinder.The experimental results show that the model can realize the prediction of the moisture content of tobacco at the exit of the drying cylinder,and the prediction error is less than 0.3%,with high prediction accuracy.The research helps to improve the quality of tobacco.

关 键 词：机器学习 烘丝筒出口 叶丝含水率 预测误差 差分进化算法 极限学习机 

分 类 号：TH-39[机械工程]