基于LSTM-MPC的粮食干燥机智能控制方法研究  

Research on Intelligent Control Method of Grain Drying Based on LSTM-MPC

作　　者：金毅 谢辉煌 尹君 张忠杰 JIN Yi;XIE Hui-huang;YIN Jun;ZHANG Zhong-jie(Institute of Grain Storage and Logistics,Academy of National Food and Strategic Reserves Administration,Beijing 100037,China;National Engineering Research Centre for Grain Storage and Logistics,Beijing 100037,China;Jiangsu Key Laboratory of Advanced Food Manufacturing Equipment and Technology,School of Mechanic Engineering,Jiangnan University,Wuxi,Jiangsu 214122,China)

机构地区：[1]国家粮食和物资储备局科学研究院粮食储运研究所,北京100037 [2]粮食储运国家工程研究中心,北京100037 [3]江南大学机械工程学院江苏省食品先进制造装备技术重点实验室,江苏无锡214122

出　　处：《粮油食品科技》2023年第5期25-34,共10页Science and Technology of Cereals,Oils and Foods

基　　金：中央级公益性科研院所基本科研业务费专项(JY2303)。

摘　　要：当前粮食干燥过程控制方法主要致力于保障粮食含水率均匀度、节能减排等,对干燥品质的精准调控尚属瓶颈问题。为应对粮食干燥过程强耦合、非线性、品质调控难等多方面问题,在前期研究的基础上,提出利用长短期记忆神经网络(LSTM)和模型预测控制(MPC)耦合控制器,结合优质稻谷品质定向调控干燥工艺参考图对粮食干燥过程进行调控。在仿真环境下对比了两种控制器的控制精度;开展了优质稻谷连续干燥实验,对比了嵌入工艺参考图前后的系统控制效果。结果表明:与常规PID控制器相比,LSTM-MPC控制器响应速度提升15%~30%,干燥后出机水分控制精度提升0.2%以上,且具有更强的鲁棒性;利用干燥工艺参考图可以实现干燥过程品质指标可视化,工艺参考图参与控制决策后,干燥后稻谷品质显著提升。The current control methods in the grain drying process mainly focus on ensuring uniform moisture content,energy saving,and emission reduction.However,precise control of drying quality remains a bottleneck problem.To address issues such as strong coupling,nonlinearity,and difficult quality control in the grain drying process,this paper proposes the utilization of long and short-term memory neural network(LSTM)coupled with model predictive control(MPC)based on previous studies.This combined controller was applied to regulate the grain drying process according to a reference diagram for high-quality rice targeted regulation during the drying process.The simulation environment was used to compare the control accuracy of these two controllers.The effectiveness of system control was evaluated through continuous rice drying tests conducted both before and after the incorporation of the process reference diagram.Results demonstrate that compared to conventional PID controller,the LSTM-MPC controller exhibits stronger robustness and faster response speed.Furthermore,by incorporating the process reference diagram into control decisions,visualization of the drying process quality index can be achieved while significantly improving overall drying quality levels.

关 键 词：优质稻谷 LSTM-MPC 品质定向调控 智能控制 

分 类 号：TS203[轻工技术与工程—食品科学] S-3[轻工技术与工程—食品科学与工程]