带钢冷轧AGC系统的数据挖掘与厚差溯源  

Data Mining and Error Tracing in the AGC System of Cold Rolling Mill

作　　者：王益群[1] 刘涛[1] 姜万录[1] 张伟[1] 

机构地区：[1]燕山大学,秦皇岛066004

出　　处：《中国机械工程》2007年第2期175-178,共4页China Mechanical Engineering

基　　金：国家自然科学基金资助项目(60474044);河北省自然科学基金资助项目(E2004000221)

摘　　要：为了获得轧制过程各扰动量的精确信息,改善冷轧带钢厚度控制(AGC)的控制效果,在AGC系统中引入了误差分析和误差溯源理论,对厚差产生的各种主要因素和环节作出了分析和描述。依托轧制过程数据库的大量数据,运用数据挖掘理论对轧制数据进行综合分析,将其中包含的主要干扰量进行逐一分离和溯源,通过回归分析确定各干扰量的性质与特征,并对回归分析得到的干扰量模型用神经网络进行校验。To acquire the exactitude information of interferences in cold milling process, and improve the control effect of AGC system, the main sources and reasons of thickness error were analyzed and described in detail using date mining theory and error tracing technology. Where the database of cold milling process was relied on as data resource, the interferences hided in it was separated and traced to sources. Through regression analysis, the characteristics of these inferences are distinguished, and the main parameter is verified by a ANN system especially designed. Then the model of inferences is made sure, which is very important to improve the precision of AGC.

关 键 词：冷轧 AGC 数据挖掘 误差溯源 人工神经网络 

分 类 号：TG331[金属学及工艺—金属压力加工]