首钢京唐1580热连轧精轧模型自学习功能及应用  

Self-learning and application of finishing rolling model for Shougang Jingtang 1 580 mm hot strip mill

作　　者：何凌云[1] 龚彩军[1] 

机构地区：[1]北京首钢自动化信息技术有限公司自动化研究所,北京100041

出　　处：《冶金自动化》2013年第6期34-37,共4页Metallurgical Industry Automation

摘　　要：首钢京唐1580热连轧生产线精轧设定模型自学习采用分钢族、分钢种、分规格和分机架的处理方式,针对主要的工艺参数建立了短期和长期自学习算法。通过中间间接变量的导出、参数估计和平滑处理等步骤进行模型参数的自学习,提高了设定模型的预报精度;通过自学习层别的细化大大改善了变钢种、变规格轧制时成品厚度及终轧温度偏差较大的现象。自学习功能投入后,运行稳定,效果良好。In Shougang Jingtang 1 580 mm hot strip mill, self-learning of setting model is processing by method of dividing steel classes,steel grades, specifications, and stands. Short and long term self-learn- ing algorithms for main process parameters are established. Through self-learning including intermediate variables derivation, parameters estimation ＇and smoothing processing, the prediction precision of setting model is enhanced. Thickness of finished products and final roiling temperature large deviation is great- ly improved by learning layer subdividing when the grade and specification are changed. Since the self- learning function has been put into use, the operation is stable, and the effect is good.

关 键 词：热连轧 过程控制系统 精轧机组 设定模型 模型自学习 

分 类 号：TG335.1[金属学及工艺—金属压力加工]