模糊神经网络融合建模方法及其在轧制力控制中的应用  

Modeling method of fuzzy neural network and its application in rolling force control

作　　者：邱华东[1] 田建艳 王书宇 菅垄 刘咸贺 韩高鹏 QIU Hua-dong;TIAN Jian-yan;WANG Shu-yu;JIAN Long;LIU Xian-he;HAN Gao-peng(Hot Continuous Rolling Plant,Taiyuan Iron and Steel Group Co.,Ltd.,Taiyuan 030003,Shanxi,China;College of Electrical and Power Engineering,Taiyuan University of Technology,Taiyuan 030024,Shanxi,China)

机构地区：[1]太原钢铁集团有限公司热轧厂,山西太原030003 [2]太原理工大学电气与动力工程学院,山西太原030024

出　　处：《中国冶金》2021年第1期52-58,共7页China Metallurgy

基　　金：山西省重点研发计划资助项目(201903D121062)。

摘　　要：针对目前热轧中神经网络控制模型不能满足一些特殊轧制规律钢种精度要求的问题,在深入研究现有热轧模型建立与优化的基础上,结合模糊控制技术,提出在神经网络的基础上建立基于模糊规则补偿模型的融合建模方法。针对两类特殊钢种的特性,详细阐述了基于模糊规则补偿模型的建立及实际应用过程,并根据实际生产经验给出建模中规则库的建立过程。实际生产过程应用结果表明,所提出的模糊神经网络融合建模方法可以有效提高轧制力计算精度和厚度控制精度,从而提高热轧带钢产品质量。In view of the problem that the neural network control model in hot rolling cannot meet the accuracy requirements of some special rolling laws, on the basis of the research of the foundation and optimization of existing hot mill model, a fusion modeling method based on the fuzzy rule compensation model neural network was proposedcombined with advanced fuzzy control technology.According to the characteristics of two types of special steels,the establishment and practical application process of compensation model based on fuzzy ruleswas explained in details.With the practical production experience,the establishment process of the rule base in the modeling process was confirmed.The application results of practical production process show that the modeling method of fuzzy neural network can effectively improve the calculation accuracy of rolling force and the control precision of thickness, thus improving the quality of hot rolling strip products.

关 键 词：热轧 轧制压力 神经网络 模糊技术 融合建模 

分 类 号：TG335[金属学及工艺—金属压力加工] TP183[自动化与计算机技术—控制理论与控制工程]