中国烟草学报2024，Vol.30Issue(2)：27-37,11.DOI:10.16472/j.chinatobacco.2023.T0082
基于模型预测控制的松散回潮自适应控制系统
An adaptive control system of tobacco loosening and conditioning based on model predictive control
摘要
Abstract
To solve the problems of poor adaptive and unstable control in traditional control of moisture regain.Based on the working principle of the loose moisture regain machine and the mechanism of moisture absorption of tobacco leaves,three key variables:the moisture content of the material inlet,the amount of water added and the return air temperature,were selected.With the outlet moisture content of the material as the target value,a prediction model was established.To address control errors caused by model deviations,using model predictive control(MPC),the backpropagation algorithm in the neural network model was used to optimize the loss function tailored to the characteristics of the moisture regain process.By enabling the predictive model to self-iterate and adapt during the control process,the adaptability of the model is improved.After system optimization,the average standard deviation of loose regain outlet moisture decreased from 0.29 to 0.20,a year-on-year decrease of 32%,and the average CPK increased from 1.132 to 1.479,a year-on-year increase of 30%.This effectively improved the control ability of loose moisture regain process.关键词
松散回潮/模型预测控制/自适应反向传播算法/损失函数Key words
tobacco loosening and conditioning/model predictive control/adaptive back propagation algorithm/loss function引用本文复制引用
赵春元,王略韬,张晓峰,于红丽,周成林,邓红伟,李秀芳..基于模型预测控制的松散回潮自适应控制系统[J].中国烟草学报,2024,30(2):27-37,11.基金项目
基于广义预测控制的松散回潮控制系统推广应用研究(20184100033450054) （20184100033450054）
