热连轧粗轧非对称板形控制模型研究及应用
发布时间：2018-10-22 13:02
【摘要】：板形理论与控制技术的研究是板带轧制领域中的热点课题,一直以来备受研究人员及生产企业的高度关注。目前,对称板形控制技术的研究已相对成熟,而由于存在着复杂的产生机理,且包含大量非线性与耦合关系的影响,非对称板形控制技术的研究却相对不足,使得长期以来没有开发出有效的在线控制策略与控制模型,生产过程只能依赖人工干预,导致控制精度不足且易于产生板形质量问题,严重影响了带钢的成品质量。本文以某2250mm常规热连轧粗轧机组为研究对象,针对热轧粗轧区板坯非对称板形问题,建立了基于多种数值计算方法的粗轧板坯镰刀弯综合调平设定模型。论文取得了如下研究成果： (1)建立了出口板坯镰刀弯弯曲量计算公式。通过建立轧辊-轧件动态热力耦合有限元模型,从来料和设备两方面考虑,分析了粗轧出口板坯非对称板形的产生机理,并验证了该计算公式的准确性,为实现粗轧板坯镰刀弯调平设定模型提供了理论基础。 (2)建立了基于两侧轧制力差的镰刀弯调平计算模型。通过弹跳方程与有限元方法相结合,针对引起板坯镰刀弯的主要因素——轧机两侧纵向刚度偏差、来料楔形及走偏分别计算对应的调整量,结合镰刀弯弯曲量公式,计算出粗轧机各道次的辊缝倾斜调整值。该模型具有完备的理论基础,是适应于粗轧机各道次的有效调整方法。 (3)建立了基于中心线偏移量的镰刀弯调平计算模型。通过测宽仪实测中心线偏移量数据解析板坯镰刀弯信息,并转化为板坯两侧厚度差,进而求解出辊缝倾斜调整值。模型中引入横向流动因子定义,并结合有限元方法,有效保证了针对不同工况的适应性。该模型能够基于镰刀弯实测数据实现粗轧机第1道次和偶数道次的有效调整。 (4)提出了综合两种调平计算模型的控制策略,建立了基于RBF神经网络的镰刀弯综合调平预报模型和在线自学习,优化各道次辊缝倾斜调整值的修正系数,能够通过实测轧制数据不断提高镰刀弯调平模型设定精度,形成了具有自动优化能力的完整镰刀弯控制体系。镰刀弯调平设定模型自投入实验运行后,中间坯镰刀弯弯曲量未达标率从24.88%下降到6.62%,表明该模型有效地控制了粗轧中间坯镰刀弯问题,提高了镰刀弯控制水平。
[Abstract]:The research of flatness theory and control technology is a hot topic in the field of strip rolling, and has been highly concerned by researchers and production enterprises. At present, the research of symmetrical shape control technology has been relatively mature, but the research of asymmetric shape control technology is relatively insufficient due to the existence of complex production mechanism and the influence of a large number of nonlinear and coupling relations. For a long time, there is no effective on-line control strategy and control model, and the production process can only rely on manual intervention, which leads to the lack of control precision and easy to produce shape quality problems, which seriously affects the quality of the finished steel strip. In this paper, aiming at the problem of asymmetric slab shape in the rough-rolled area of 2250mm, a comprehensive leveling model for sickle bending of rough-rolled slab is established based on various numerical calculation methods. The main results are as follows: (1) the calculation formula of bending quantity of sickle for export slab is established. By establishing the dynamic thermo-mechanical coupling finite element model of roll and workpiece, considering the material and equipment, the mechanism of asymmetric slab shape at the outlet of rough rolling is analyzed, and the accuracy of the calculation formula is verified. This paper provides a theoretical basis for the setting model of sickle bending leveling for rough rolling slab. (2) the calculation model of sickle bending leveling based on the rolling force difference between two sides is established. Through the combination of spring equation and finite element method, aiming at the main factors that cause slab sickle bending, such as longitudinal stiffness deviation of both sides of rolling mill, feed wedge and walking deviation, the corresponding adjustment amount is calculated respectively, and combined with the formula of sickle bending amount, The roll slant adjustment value of each pass of roughing mill is calculated. This model has a complete theoretical foundation and is an effective adjustment method for each pass of roughing mill. (3) the calculation model of sickle bending leveling based on centerline offset is established. The deflection data of the center line measured by the width meter are used to analyze the bending information of the sickle of the slab, and the thickness difference between the two sides of the slab is transformed into the thickness difference between the two sides of the slab, and then the roll slanting adjustment value is solved. The definition of transverse flow factor and the finite element method are introduced into the model to ensure the adaptability of different working conditions. Based on the measured data of sickle bend, the model can effectively adjust the first pass and even number pass of roughing mill. (4) the control strategy of synthesizing two kinds of leveling calculation models is proposed. A comprehensive leveling prediction model of sickle bending based on RBF neural network and on-line self-learning are established to optimize the correction coefficient of roll slant adjustment value for each pass, which can continuously improve the setting accuracy of sickle bending leveling model through the measured rolling data. A complete sickle bending control system with automatic optimization capability is formed. After the sickle leveling model has been put into operation since the experiment, the bending rate of the sickle is decreased from 24.88% to 6.62%, which indicates that the model can effectively control the bending problem of the sickle and improve the control level of the sickle bending.
【学位授予单位】：北京科技大学
【学位级别】：博士
【学位授予年份】：2015
【分类号】：TG334.9
本文编号：2287232
[Abstract]:The research of flatness theory and control technology is a hot topic in the field of strip rolling, and has been highly concerned by researchers and production enterprises. At present, the research of symmetrical shape control technology has been relatively mature, but the research of asymmetric shape control technology is relatively insufficient due to the existence of complex production mechanism and the influence of a large number of nonlinear and coupling relations. For a long time, there is no effective on-line control strategy and control model, and the production process can only rely on manual intervention, which leads to the lack of control precision and easy to produce shape quality problems, which seriously affects the quality of the finished steel strip. In this paper, aiming at the problem of asymmetric slab shape in the rough-rolled area of 2250mm, a comprehensive leveling model for sickle bending of rough-rolled slab is established based on various numerical calculation methods. The main results are as follows: (1) the calculation formula of bending quantity of sickle for export slab is established. By establishing the dynamic thermo-mechanical coupling finite element model of roll and workpiece, considering the material and equipment, the mechanism of asymmetric slab shape at the outlet of rough rolling is analyzed, and the accuracy of the calculation formula is verified. This paper provides a theoretical basis for the setting model of sickle bending leveling for rough rolling slab. (2) the calculation model of sickle bending leveling based on the rolling force difference between two sides is established. Through the combination of spring equation and finite element method, aiming at the main factors that cause slab sickle bending, such as longitudinal stiffness deviation of both sides of rolling mill, feed wedge and walking deviation, the corresponding adjustment amount is calculated respectively, and combined with the formula of sickle bending amount, The roll slant adjustment value of each pass of roughing mill is calculated. This model has a complete theoretical foundation and is an effective adjustment method for each pass of roughing mill. (3) the calculation model of sickle bending leveling based on centerline offset is established. The deflection data of the center line measured by the width meter are used to analyze the bending information of the sickle of the slab, and the thickness difference between the two sides of the slab is transformed into the thickness difference between the two sides of the slab, and then the roll slanting adjustment value is solved. The definition of transverse flow factor and the finite element method are introduced into the model to ensure the adaptability of different working conditions. Based on the measured data of sickle bend, the model can effectively adjust the first pass and even number pass of roughing mill. (4) the control strategy of synthesizing two kinds of leveling calculation models is proposed. A comprehensive leveling prediction model of sickle bending based on RBF neural network and on-line self-learning are established to optimize the correction coefficient of roll slant adjustment value for each pass, which can continuously improve the setting accuracy of sickle bending leveling model through the measured rolling data. A complete sickle bending control system with automatic optimization capability is formed. After the sickle leveling model has been put into operation since the experiment, the bending rate of the sickle is decreased from 24.88% to 6.62%, which indicates that the model can effectively control the bending problem of the sickle and improve the control level of the sickle bending.
【学位授予单位】：北京科技大学
【学位级别】：博士
【学位授予年份】：2015
【分类号】：TG334.9
【参考文献】
相关期刊论文 前10条
1 李林;;SP定宽轧制板形侧弯机理与策略[J];宝钢技术;2007年02期
2 吴刚,孙一康;多变量模糊预测控制在板形板厚综合系统中的应用[J];北京科技大学学报;1999年04期
3 王爱丽;杨荃;何安瑞;刘华强;;热连轧粗轧区FES宽展模型及其优化[J];北京科技大学学报;2010年04期
4 宋勇;苏岚;荆丰伟;刘文仲;;热轧带钢轧制力模型自学习算法优化[J];北京科技大学学报;2010年06期
5 杨荃，陈先霖;轧制带材的瓢曲生成路径[J];北京科技大学学报;1994年01期
6 费庆,战守义,胡浩平,张迪生;基于神经网络的热轧带钢宽度预报与设定[J];北京理工大学学报;2004年12期
7 张殿华,王君,郑芳,王国栋;板带热连轧机组活套高度模糊控制[J];东北大学学报;2000年02期
8 胡贤磊,王昭东,于解民,刘相华;结合模型自学习的BP神经元网络的轧制力预报[J];东北大学学报;2002年11期
9 赵宪明,王国栋,朴海斗,崔承甲;立辊形状对粗轧板坯侧弯的影响[J];东北大学学报;2002年12期
10 丁敬国;焦景民;昝培;刘相华;;基于模糊聚类的PSO-神经网络预测热连轧粗轧宽度[J];东北大学学报(自然科学版);2007年09期
，本文编号：2287232
本文链接：https://www.wllwen.com/kejilunwen/jinshugongy/2287232.html
教材专著
