中厚板轧机轧制力自学习模型的应用

• 作 者: 许磊 曾庆亮 胡贤磊 王君 王国栋
• 机 构: 东北大学轧制技术及连轧自动化国家重点实验室,安阳钢铁集团公司,
• 出 处: 钢铁研究学报
• 发表日期: 2009年12期
• 关键词: 学习模型 中厚板轧机 层别划分 开轧温度 学习系数 轧制力 出口厚度 在线应用 预报精度 自学习

针对传统轧制力模型在现场应用中预报精度不高的现状,利用轧制过程中的仪表反馈数据开发出一种新的轧制力自学习模型。对计算所得轧制力自学习系数按照钢种、道次出口厚度、开轧温度、目标厚度、目标宽度等要素进行了层别的划分,并对其依据和效果分别做了系统的分析。现场在线应用结果表明,给出的轧制力模型具有良好的预测精度,末道次轧制力预测误差可以控制在3%以内,其它道次可以控制在5%以内。

Aiming at the current situation that the traditional rolling force model is not accurate in the field application, a new rolling force self-learning model is developed by using the instrument feedback data in the rolling process. The self-learning coefficient of the calculated rolling force is divided into layers according to the factors such as the type of steel, the exit thickness of the pass, the rolling temperature, the target thickness and the target width. The basis and effect of the self-learning coefficient are systematically analyzed respectively. The on-line application results show that the proposed rolling force model has a good prediction accuracy, the prediction error of the last pass rolling force can be controlled within 3%, and the other pass can be controlled within 5%.