冷轧带钢预设定过程板形板厚耦合特性研究及协调优化

There is an obvious coupling relationship between flatness and gauge when process presetting of thin strip in tandem cold rolling. Conventional methods often study them separately, making it difficult to obtain the ideal control effect. How to improve the control accuracy of flatness preset process and the quality of strip head shape is the key to further improve the control level of tandem cold rolling and the problem to be solved urgently. The big data and intelligent control are combined with rolling mechanism to drive the bending force preset optimization. The integration and storage of steel big data are realized based on clustering analysis and database technology. The common law of influence that rolling force and bending force on flatness are explored by data mining algorithm and the accurate prediction of strip flatness is realized by support vector machine optimized by particle swarm optimization. The law that optimum bending force changes with rolling force is revealed by fuzzy reasoning and the output parameters of flatness control are predicted through neural network optimized by genetic algorithm. Based on the above research, the coupling characteristics between rolling force and bending force are studied. The intelligent control strategy of flatness preset is proposed and the new morphological model of bending force preset is built by multi-objective optimization theory to achieve the decoupling control and coordinated optimization of flatness and gauge, then the intelligent solution of model optimization is given. The research results will provide new ideas for improving the preset accuracy of thin strip flatness and the intelligent preparation of steel in the whole process.

冷连轧薄带材的板形板厚在工艺预设定时存在明显的耦合关系，常规方法多将其分开来单独研究，难以获得理想的控制效果，如何提高板形预设定过程的控制精度和带钢头部板形质量，是冷连轧板形控制水平进一步提升的关键和亟待解决的问题。本项目以轧制机理融合大数据和智能控制来驱动弯辊力预设定优化，采用聚类分析和数据库技术实现钢铁大数据的集成和存储；基于数据挖掘算法探寻轧制力及弯辊力对板形的共性影响规律，采用粒子群算法优化的支持向量机实现带钢板形的精准预测；通过模糊推理揭示最优弯辊力随单位宽度轧制力的变化规律，采用遗传算法优化的神经网络预报板形控制输出参数，实现板形板厚耦合特性的研究；制定板形预设定智能控制策略，采用多目标优化理论构建弯辊力预设定新形态模型，实现板形板厚解耦控制及协调优化，给出模型寻优的智能求解方法。研究成果可为冷连轧薄带材板形预设定精度的提高和钢铁全流程智能制备提供新的思路。

本研究以板形预设定阶段板形板厚的解耦控制为研究思路，通过先进控制理念及算法改善带钢头部板形控制精度，提出一种基于目标函数的冷连轧轧制力和前滑模型协同自适应算法，提高轧制力模型和前滑模型的设定精度；开展弯辊对带钢板形的控制特性分析，揭示不同板形执行机构的板形调控功效及对接触变形区轧制压力的影响规律；提出兼顾轧制力的弯辊力预设定控制策略，开展基于大数据的板形预设定模型研究，构建弯辊力预设定新形态模型；针对冷连轧带钢出口板形预测提出一种基于小波神经网络的集成学习方法，提高带钢出口板形预测精度；开发板形预设定控制系统模块，实现非稳定轧制阶段的带钢头部板形标准差＜14I，带钢全长板形合格率100%，实现冷连轧板形预设定新形态模型的在线应用，提高板形控制精度。研究成果有望推动钢铁全流程智能制备关键技术发展，加快新一代信息技术与钢铁工业深度融合，为钢铁全流程智能制造系统的建立和运行奠定基础，将对早日实现钢铁轧制过程全流程智能化控制起到重要推动作用。