基于神经网络的烟叶复烤回潮过程预测控制模型构建

Construction of Predictive Control Model for Moisture Regain Process of Tobacco Leaf Redrying Based on Neural Network

针对烟叶复烤回潮过程系统复杂、参数耦合干扰、时间滞后等问题,以两段式烟叶复烤回潮工艺为研究对象,依据生产中工艺参数实时数据,构建了基于支持向量机(SVM)神经网络的烟叶复烤回潮过程预测控制模型。该模型通过训练学习,可以准确描述出口含水率与其他工艺参数之间的内在关系,有效预测二次润叶出口含水率,最大绝对误差不超过0.45%,为烟叶复烤回潮工艺提供了可行的预测控制方案,有利于出口含水率的有效调控,对烟丝生产加工具有重要的指导意义。

In response to the problems of complex system,parameter coupling interference and time delay during the moisture regain process of tobacco leaf redrying,based on real-time data of production process parameters,a predictive control model for the moisture regain process of tobacco leaf redrying based on support vector machine(SVM)neural network was constructed using the moisture regain process of two-stage tobacco leaf redrying as the research object.Through training and learning,this model could accurately describe the internal relationship between the outlet moisture content and other process parameters,effectively predict the outlet moisture content of secondary leaf moistening,with a maximum absolute error≤0.45%.It provides a feasible predictive control scheme for the moisture regain process of tobacco leaf redrying,is beneficial to the effective regulation of the outlet moisture content,and has important guiding significance for the cut tobacco production and processing.