基于DRNN-PID的板形板厚解耦控制  

Strip Flatness and Gauge Decoupling Control Based on Self-Tuning PID With Diagonal Recurrent Neural Network

作　　者：杨景明[1] 杜鹏[1] 车海军[1] 张林浩[1] 

机构地区：[1]燕山大学工业计算机控制工程河北省重点实验室,河北秦皇岛066004

出　　处：《钢铁》2010年第3期57-61,共5页Iron and Steel

基　　金："十一五"国家科技支撑计划项目(2007BAF02B12)

摘　　要：针对板形板厚综合系统具有强耦合、非线性、含纯滞后环节的特点,建立了板形板厚耦合模型,并在对其进行神经网络解耦设计的基础上提出了基于对角递归神经网络(DRNN)整定PID的板形板厚解耦控制方法,然后根据带钢冷轧情况提出神经网络解耦对不同塑性刚度参数的实际适用范围。仿真结果表明,该解耦控制系统具有比传统前馈补偿解耦PID控制效果好、响应速度快、自适应跟随能力强等优点,并且符合实际轧制要求,有效地提高了板形板厚的控制精度。In view of the process of automatic flatness control and automatic gauge control which is a nonlinear system with strong coupling and pure time delay,a strip flatness and gauge coupling model is established,then a strip flatness and gauge decoupling control method is developed based on self-tuning PID with diagonal recurrent neural network（DRNN） on the basis of the neural network decoupling design,and a scope is available for the different plastic rigidity parameter.The simulation results indicate that control system has the good decoupling effect,and the speed of response is quick,and has good performances of adaptively tracking target and so on.Meanwhile it meets the actual rolling requirements and enhances effectively strip flatness and gauge control precision.

关 键 词：解耦 神经网络解耦 对角递归神经网络 板形板厚 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]