步进式加热炉过程控制模型研究

发布时间：2018-04-20 16:07

本文选题：加热炉 + 过程控制　；参考：《北京科技大学》2017年博士论文

【摘要】：加热炉是连接连铸和轧线的关键中间设备,用于加热板坯使之达到轧制温度。作为加热炉的核心控制系统,过程控制系统的主要任务是根据生产工艺和相关数学模型来控制和协调生产设备,通过优化设定获得符合轧制温度要求的板坯。由于板坯加热过程具有时间长、变量多和非线性等特点,对其温度的预报较为复杂,为了建立兼顾精确性和快速性的在线模型,需要在众多参数(辐射、对流、板坯灰度、氧化层、水梁等)中选取对其影响较大的因素。同时,轧制计划的编排很难避免板坯混装的情况出现,这种混装导致的板坯材料和规格的复杂多变性,使得现有研究中对板坯进行批量处理的控制方法较为粗糙。本文针对这些问题,在热传导机理模型的基础上,采用数值建模的方法,构建了改进的板坯温度预报模型和加热炉炉温在线设定模型。取得了如下创新性成果:(1)通过建立板坯三维温度场模型,定量分析了氧化层和水梁"黑印"对板坯温度计算的影响,并得到板坯在长宽高方向的温度分布。在此基础上,引入氧化烧损模型,在"黑印"上方、板坯长高方向上选取计算域,提出了一种考虑氧化层增长的、适于在线应用的板坯二维温度预报模型,并通过埋偶实验获得的板坯表面实测值,验证了该模型的精度。(2)为校正板坯内部温度的预报偏差,提出板坯内部温度预报校正模型。该模型考虑了板坯温度预报模型中参数的不确定性,并用PCA模型表示这种不确定性下的模型偏差;在对计算域中实验点的单元格进行偏差校正后,基于MLS建立单元格参数与模型偏差的响应曲面,以实现利用有限的实验数据对计算域中温度偏差进行近似与校正的功能。实验结果显示,校正后的偏差值比校正前下降了 42%,有效弥补了原始模型的偏差。(3)考虑不同钢种、规格板坯混装的实际复杂工业生产情况,建立加热炉炉温在线设定模型。该模型利用基于改进遗传算法的加热炉炉温离线优化模型所得到的炉温值,考虑钢种等级、板坯厚度等四个评价指标,采用改进型熵权-TOPSIS法对每块板坯对应的炉温进行赋权,以得到兼顾所有板坯属性的最优炉温设定值,提高了板坯温度的控制精度。将本系统应用于某钢厂加热炉后,解决了其原有模型预报精度低,能耗高等难题;大幅度提高了 RDT预报精度,减少了人工操作时间,在保证加热质量的情况下降低了炉温;为厂方提高生产效率的同时节约了能源。
[Abstract]:Heating furnace is the key intermediate equipment to connect continuous casting and rolling line. It is used to heat slab to reach rolling temperature. As the core control system of the reheating furnace, the main task of the process control system is to control and coordinate the production equipment according to the production process and related mathematical models. Because the slab heating process has the characteristics of long time, many variables and nonlinear, the prediction of the temperature is more complicated. In order to establish the on-line model with both accuracy and rapidity, many parameters (radiation, convection, slab gray scale) are needed. The oxidation layer, water beam, etc. At the same time, the rolling schedule arrangement is difficult to avoid the situation of slab mixed loading, which leads to the complex variability of slab material and specification, which makes the control method of batch processing of slab in the existing research rough. In this paper, based on the heat conduction mechanism model, an improved slab temperature prediction model and a furnace temperature on-line setting model are constructed by numerical modeling. Through the establishment of three-dimensional temperature field model of slab, the influence of oxide layer and "black print" of water beam on slab temperature calculation is quantitatively analyzed, and the temperature distribution of slab in the direction of length, width and height is obtained. On the basis of this, the oxidation burn model is introduced, and the calculation field is selected in the direction of slab length and height above "black print". A two-dimensional temperature prediction model of slab is proposed, which is suitable for on-line application and takes account of the growth of oxide layer. The accuracy of the model is verified by the experimental data obtained from the buried couple experiment, which is used to correct the prediction deviation of the internal temperature of the slab, and a correction model for the prediction of the internal temperature of the slab is put forward. The model takes into account the uncertainty of parameters in the slab temperature prediction model, and uses PCA model to express the model deviation under this uncertainty. The response surface of cell parameter and model deviation is established based on MLS to realize the function of approximating and correcting the temperature deviation in computational domain using limited experimental data. The experimental results show that the deviation value after correction is 42% lower than that before correction, which effectively makes up for the deviation of the original model. (3) considering the actual complex industrial production situation of different steel grades and mixed slab, the on-line setting model of furnace temperature for heating furnace is established. The model uses the furnace temperature value obtained from the off-line optimization model of furnace temperature based on improved genetic algorithm, considering four evaluation indexes such as steel grade and slab thickness, and uses the improved entropy weight TOPSIS method to weight the furnace temperature corresponding to each slab. In order to obtain the optimal furnace temperature setting value which takes all slab attributes into account, the control precision of slab temperature is improved. The system has been applied to a reheating furnace in a steel plant to solve the problems of low prediction precision and high energy consumption of the original model, greatly improve the precision of RDT prediction, reduce the manual operation time, and reduce the furnace temperature under the condition of ensuring the heating quality. Improve the production efficiency for the factory at the same time saving energy.
【学位授予单位】：北京科技大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：TG307;TP273

【参考文献】