基于时滞神经网络的铝带冷轧板形板厚解耦控制  

Flatness and thickness decoupling control of aluminum cold rolling mill using delayed neural network

作　　者：张潮[1] 郭强[1] 王京[1] 

机构地区：[1]北京科技大学高效轧制国家工程研究中心,北京100083

出　　处：《内蒙古科技大学学报》2014年第4期343-348,共6页Journal of Inner Mongolia University of Science and Technology

基　　金：北京市青年英才计划基金资助项目(YETP0412)

摘　　要：在铝带冷轧过程中,板形和板厚两个指标之间存在复杂的耦合关系,对于板形通道单独进行单独控制的方法会对厚度通道造成影响,反之亦然.在某些工况下,甚至会造成整个系统的不稳定.本文在建立单机架冷轧板形板厚耦合机理模型的基础上,利用时滞神经网络结合PID控制器建立板形板厚综合解耦控制系统.根据宁夏某铝带生产现场数据进行仿真研究,仿真结果验证了本文方法的有效性和鲁棒性.In aluminum cold rolling process, flatness and thickness control systems interact each other intensively. Adjusting one of these two factors will bring disturbance to another, and even make the system unstable sometimes. The flatness and thickness dynamic coupling model was built, and the PID controller based on delayed neural network was used to deeouple the MIMO system to weaken the coupling effects. Simulation of the field data from one factory in Ningxia Province validates the effectiveness of the algorithm for the coupling system.

关 键 词：铝带 板形控制 厚度控制 解耦控制 神经网络 

分 类 号：TG335.5[金属学及工艺—金属压力加工]