UC轧机二次型板形缺陷模糊神经网络控制  

Fuzzy Neural Network Control for Shape Defects of Quadratic Form of UC Rolling Mill

作　　者：陈丽[1] 李建更[1] 乔俊飞[1] 王笑波[2] 

机构地区：[1]北京工业大学电子信息与控制工程学院,北京100022 [2]宝钢技术中心自动化所,上海201900

出　　处：《控制工程》2005年第5期438-441,共4页Control Engineering of China

基　　金：国家自然科学基金与宝钢集团公司联合资助项目(50274003)

摘　　要：针对UC轧机中间辊弯辊回路的非线性过程精确控制问题,将模糊控制与神经网络结合,设计模糊神经网络控制器并将其应用于对中间辊弯辊回路的控制中。介绍了由神经网络结构组成的模糊控制器,网络完成模糊化、模糊推理、清晰化操作,模糊规则和系统的输入输出由梯度下降法进行调整。仿真结果表明,这种模糊神经网络控制器能很好地跟踪二次板形的目标设定值,系统的响应快,超调小,鲁棒性强。The problem of shape defects of quadratic form in the control system of universal crown rolling mill is discussed. To realize accurate control of bending of the intermediate mils, a fuzzy neural network is presented to control the loops of bending of the intermediate mils in which fuzzy logic control and neural network are combined. The experimental result shows that the control effect is excellent, with small ultra-regulated and good robustness, reaching the goal of the intelligence control for shape on line.

关 键 词：板形控制 液压弯辊 模糊神经网络 BP网络 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]