首页|基于BP神经网络的轧机刚度预测模型

基于BP神经网络的轧机刚度预测模型

Prediction model of rolling mill stiffness based on BP neural network

扫码查看
原文链接
  • NETL
  • NSTL
  • 万方数据
轧机刚度对热轧带钢厚度、板形控制等有重要影响,轧机牌坊与辊系轴承座之间间隙变化引起的辊系轴线交叉是导致轧机刚度变化的主要因素.为此,以某精轧机为研究对象,建立了轧机的数值仿真模型,并以轧机设计刚度对模型正确性进行验证.基于该仿真模型,采用正交试验方法,通过修改轧机牌坊与辊系轴承座之间的间隙,计算不同辊系轴线交叉状态下的轧机刚度值,构建了训练数据集.采用BP神经网络建立间隙与轧机刚度之间的非线性映射数学模型,实现了轧机刚度预测.现场的刚度试验和轧机刚度调整实践均验证了该模型的可靠性.该模型为轧机刚度精确预测、轧机牌坊和辊系轴承座间隙调整以及轧机精度智能调整提供了理论依据.
The rolling mill stiffness is an important parameter for the thickness and shape control of hot rolled strip.The change of the cross state of the roll system axis caused by the change of the gaps between the bearings of roll system and the rolling mill housing is the main factor causing the fluctuation of the rolling mill stiffness.Therefore,a finishing rolling mill was used to re-search the rolling mill stiffness prediction model.The FEA model of finishing rolling mill was built firstly,and the accuracy of the FEA model was verified by comparison with the design stiffness.Based on the FEA model,an orthogonal simulation test was implemented,the rolling mill stiffness values under different cross state of roll system axis were calculated by modifying the gaps between the rolling mill housing and the bearings of roll system,and the training dataset was constructed.Then the Back Propaga-tion Neural Network(BPNN)was applied to fit the nonlinear mapping relationship between the gap and rolling mill stiffness for the pre-diction of rolling mill stiffness.The reliability of prediction model was verified by stiffness test and the practice of stiffness adjusting.The prediction model provides a theoretical basis for the accurate prediction of rolling mill stiffness,the adjustment of the gap between the rolling mill housing and the bearings of roll system,and the intelligent adjustment of the rolling mill precision.

rolling mill stiffnesscross state of roll systemBP neural networkgap between the bearing of roll system and the rolling mill housingstiffness prediction modelintelligent adjustment

邱碧涛、但斌斌、肖涵、阮金华、袁锐

展开 >

武汉科技大学冶金装备及其控制教育部重点实验室,湖北 武汉 430081

武汉钢铁有限技术中心,湖北 武汉 430082

轧机刚度 辊系交叉 BP神经网络 牌坊与辊系轴承座之间间隙 刚度预测模型 智能调整

国家自然科学基金项目

51701145

2024

10.13228/j.boyuan.issn1003-A9996.20240216

轧钢
中国钢研科技集团有限公司

轧钢

CSTPCD北大核心
影响因子：0.881
ISSN：1003-9996
年,卷(期)：2024.41(2)
  • 22