【作者】 曹光明；

【导师】 吴迪； 刘振宇；

【作者基本信息】 东北大学 ， 材料加工工程， 2008， 博士

【摘要】 近年来,双辊铸轧薄带钢技术作为钢铁工业发展方向的一项新技术,得到了世界钢铁界的广泛重视,但仍需要不断完善以尽快实现工业化生产。在我国虽然双辊铸轧薄带钢技术已经研究很多年,但与国外先进技术水平存在很大的差距。在双辊铸轧薄带钢过程中,液态钢水在很短时间内完成从凝固到成形的过程,工艺参数的可控范围非常窄,工艺参数的微小变化会导致铸带出现严重缺陷,甚至出现轧卡或漏钢事故。所以在双辊铸轧薄带钢过程中若不对相关的工艺参数进行检测和控制,很难保持工艺的稳定,更加不会铸出质量良好的薄带。因此高精度的检测与控制技术是双辊铸轧薄带钢技术能否真正走向产业化的关键。本文对双辊铸轧薄带钢过程中监测系统、熔池液位控制方法、铸轧力模型开发以及工艺参数优化方法进行了研究。研究结果对于双辊铸轧机的过程控制系统的建立具有重要的理论意义和实用价值。本文首先分析和研究了双辊铸轧薄带钢工艺参数的检测方法,然后利用西门子工控设备,并结合相关的检测仪表和设备,建立了双辊薄带铸轧机工艺参数监测系统。并对实验数据进行了系统分析,从而为研究双辊铸轧薄带钢的规律性提供可靠的数据。本文的创新点和主要研究内容如下：1.在分析了熔池钢水液位系统干扰因素和控制难点的基础上,建立了熔池液位被控对象模型。研究采用主副回路PID控制器、在副回路提出采用微分先行PID控制器控制塞棒液压伺服系统,减少了塞棒的频繁移动。主回路上给出一种适用于熔池液位控制的模糊自适应控制器的设计方法。仿真结果表明,采用上述控制策略时系统响应速度快、鲁棒性强。2.根据铸轧实验、通过热平衡计算和模拟相结合的反向方法分段建立了凝固过程中凝壳与铸辊之间热传导系数与铸轧速度、熔池位置之间的关系模型,并根据三维热流耦合有限元分析理论对双辊铸轧薄带钢凝固过程中流场和温度场进行了数值模拟。给出了浇注温度、铸轧速度、熔池液位高度等参数对熔池内流场和温度场的影响规律,为铸轧工艺参数的确定奠定一定的理论基础,并在此基础上提出了一种凝固终点位置控制策略。3.在对熔池内金属凝固和变形的机理分析的基础上,以凝固终点为分界点将铸轧区域分为两个求解区域。在液相区、糊状区引入流体力学中Navier-Stokes方程和流函数,分析其流变特性并推导出单位压力分布解析式,在固相区仍沿用热轧模型。在凝固终点位置计算模型中,采用基于贝叶斯方法的前向训练算法建立凝固终点位置计算模型,避免了网络过训练,提高网络的泛化能力。实验表明将上述模型应用于铸轧力计算,具有很高的计算精度。4.针对双辊铸轧薄带钢过程中凝固终点位置这一关键参数,在上述建立的凝固终点位置计算模型和化学成份、工艺约束已知的条件下,采用粒子群优化算法对双辊铸轧薄带钢过程中相应的工艺参数进行优化,通过调整相应参数到优化值以达到稳定凝固终点位置的目的,从而为提高双辊铸轧板形和板厚的控制精度,改善铸带表面质量提供条件,同时优化结果得到实验室铸轧实验的验证。更多还原

【Abstract】 In recent years, twin-roll casting process of steel strip, as a new technology representing the development direction of steel industry, attracts the world-wide attention of the steel researchers. Improvement is still in great need to realize the industrial application. This technology has been investigated for many years in China. However, large gaps are existed between foreign and domestic researches. In the twin-roll casting process, solidification of molten metal is completed rapidly and the process window is in a narrow range. Small variation of the process parameters lead to severe defects of the casting strip, even the stop of work roll and leak of molten steel. Therefore, the measurement and control of process parameters are required to stabilize the casting process, to produce casting strip with good quality and to realize the industrial application. In the present paper, the monitoring system, melt steel level control, rolling force model and processing parameters optimization of twin-roll steel strip casting process are investigated, which provide theoretically and practically basis for establishment of process control system.The measuring methods of process parameters were first analyzed and investigated. The monitoring system of twin-roll strip casting was established by the employment of Siemens IPC and related instruments. The experimental data was analyzed systematically, which provide accurate database for study of twin-roll steel strip casting. Some chief original work and conclusions are as follows.1. The model of controlled device was established based on the analysis of interfering factors and difficulties of melt steel level system. Major and minor loop PID control was investigated and differential first PID control has been employed at minor loop to reduce frequent movement of stopper. In the major loop the fuzzy self-adaptive PID control strategy was proposed. Simulation results indicate the rapid response and robustness of the system.2. According to the experiment results, the relationship between casting speed, location and heat transfer coefficient in different region were established using the inverse method combining the heat balance calculation and simulation. Based on three dimensional finite element coupled analysis theory, a numerical simulation for flow and temperature fields during the solidification of twin-roll steel strip process was conducted. The effects of the pouring temperature, casting speed, liquid surface height in molten pool on the flow field and temperature field have been obtained to provide theoretical basis for the determination of the processing parameters. And a strategy of position of kisspoint was proposed.3. Based on analysis of solidification and deformaion mechanism in the molten pool, twin-roll casting can be divided into two zones by kiss point position. Navier-Stokes equation in fluid mechanics and stream function were introduced to analyze the rheological properities of liquid zone, mushy zone and deduce the analytic equation of average unit compression stress distribution, In the solid zone, the traditional hot rolling model was still used. In calculation of kiss point position, Bayesian regularization was applied to the training of feed forward neural networks to avoid "over-fitting" and improve the generalization ability. The models mentioned above were applied to the calculation of rolling force, which has exhibited high precision.4. In order to improve the flat, gage control precision and surface quality of casting strip, optimization of processing parameters based on kiss point position calculation model, was conducted using the particle swarm optimization method under the condition that the chemical composition and process constraints were set. The result shows that the optimized results were in good agreements with experimental one.更多还原