热轧精轧轧辊摩擦和磨损研究  

Investigation on friction and wear of finishing roll of hot rolling

作　　者：吴进[1] 邱春林[1] 齐克敏[1] 张国河[2] 献智[2] 李欣波[2] 

机构地区：[1]东北大学轧制技术与连轧自动化国家重点实验室,辽宁沈阳110004 [2]宝钢集团上海梅山钢铁有限公司,江苏南京210039

出　　处：《钢铁研究》2006年第6期31-34,共4页Research on Iron and Steel

摘　　要：针对某厂热连轧机采用定值摩擦系数计算轧制力与实际值相差2000～3000kN的情况，采用BP神经元网络与NeuroShe112软件对摩擦系数进行预测与检验，建立了符合现场实际的摩擦系数模型，在实际应用中提高了轧制力的预报精度。并根据POMINI公司磨床的辊型曲线。分析了精轧轧辊磨损的基本规律，提出了减小轧辊磨损的具体措施。Since a constant friction coefficient was used in the original control model of a hot continuous rolling mill , an error of 2 000-3 000kN in the rolling pressure prediction was caused, In this article, BP neural network and NeuroShell2 software were employed in forecasting and verifying the friction coefficient. A new friction coefficient model, which agreed with the production practice quite well, was developed and the rolling pressure prediction accuracy was highly improved. In response to roll curve attained from POMINI, basic regularities of finishing roll wear are analyzed. Concrete measures are provided to reduce roll wear.

关 键 词：摩擦系数 人工神经网络 BP算法 磨损 

分 类 号：TG333.1[金属学及工艺—金属压力加工]