摘要
针对河钢邯钢二冷轧厂1#镀锌机组气刀人工控制所具有的大滞后性、多变量性及锌层命中率低的问题,开发了基于BP神经网络的气刀闭环控制系统,对比分析了BP神经网络模型和多元线性回归模型对锌层重量的预测精度。经过合理设计并添加多种自定义选项的气刀闭环控制系统,实现了气刀压力、气刀间距及气刀高度随产线速度和目标锌层变化进行自动调整,具有高度的鲁棒性、适应性及灵活性。系统投入运行后,产品镀层厚度控制精度和成材率大幅提高。
Aiming at the problems of serious lag, multivariate and low hit rate of zinc layer of air knife manual control of 1# galvanizing mill unit in Hansteel No.1 cold rolling mill, a closed-loop control system of air knife based on BP neural network was developed, and the prediction accuracy of zinc layer weight by BP neural network model and multivariate linear regression model was compared and analyzed. The closed-loop control system of air knife with various self-defined options can achieve the automatically adjustment of pressure, spacing and height with the change of production line speed and target zinc layer. It has high robustness, adaptability and flexibility. After the system is put into operation, the control precision and yield of the coating thickness are greatly improved.
作者
宋志超
Song Zhichao(Technology Center of HBIS Group Hansteel Company, Handan, Hebei, 056015)
出处
《河北冶金》
2019年第10期26-29,65,共5页
Hebei Metallurgy
关键词
连续热镀锌
锌层厚度
BP神经网络
气刀闭环控制
精度
气刀压力
气刀间距
continuous hot dip galvanizing
zinc layer thickness
BP neural network
air knife closed-loop control
accuracy
air knife pressure
air knife spacing
