转炉炼钢终点碳温预测与控制模型研究

发布时间：2018-03-28 13:34

本文选题：转炉炼钢　切入点：终点碳温　出处：《天津理工大学》2017年硕士论文

【摘要】：我国是传统的炼钢大国,钢铁产业是重要的国民经济支柱。“十二五”发展规划以来,我国大力提倡运用新技术、新设备和新方法,降低钢铁成本,提高钢铁质量,增大我国钢铁在世界上的竞争力,让我国成为全球钢铁强国。当前,我国中小型转炉技术设备落后,迄今依然还有一部分转炉停留在人工经验判断冶炼终点的水平并且受转炉厂房环境和经济条件限制,先进的副枪检测技术并不能实地使用。在上述背景环境下,开发实用、高效且价格低廉的新技术变得意义重大。本文围绕转炉终点碳温预测控制新方法展开。首先分析了转炉内的物料平衡和热平衡,依此平衡建立静态配料计算模型,实现对配料环节物料配比的良好管理,提高了出钢的精度。其次采用处理非线性问题能力强的极限学习机对转炉终点进行预测,对于极限学习机的权值阈值随机确定导致网络结构不稳定的问题,使用遗传算法对其权值和阈值进行优化,并对极限学习机隐含层激励函数和隐含层节点数选取不同对预测结果的影响做了比较,仿真出来的误差数据和预测图形说明了该改进方法有效,实现了预测精度的提升。针对静态预测容易在有扰动的环境下发生预测不准的情况,采用粒子群算法建立基于预测的优化控制模型,为提升粒子群算法的搜索能力,对其惯性权重进行了改进,对其采用不同的函数和公式对优化结果的影响做出了比较,仿真结果验证了该策略有效性,实现了对影响终点的主控变量的优化控制。最后使用VC++搭建软件平台,将配料计算模型、预测模型和优化控制模型编制成实际操作界面,实现了钢厂工作人员对炼钢过程的预测和控制。
[Abstract]:China is a traditional steelmaking country, and the iron and steel industry is an important pillar of the national economy. Since the 12th Five-Year Plan, our country has vigorously advocated the use of new technologies, new equipment and new methods to reduce the cost of iron and steel and improve the quality of iron and steel. To increase the competitiveness of China's iron and steel in the world, and to make China a global steel power. At present, China's small and medium-sized converters are lagging behind in technology and equipment. Up to now, there are still some converters that remain at the level of the end point of smelting by artificial experience and are limited by the environment and economic conditions of the converter plant. The advanced auxiliary lance detection technology cannot be used in the field. Under the above background environment, the development is practical. The new technology with high efficiency and low cost has become significant. This paper focuses on the new method of carbon temperature prediction and control at the end of converter. Firstly, the material balance and heat balance in converter are analyzed, and the static proportioning calculation model is established according to this balance. In order to improve the precision of steel production, we can achieve good management of proportioning material ratio. Secondly, the ultimate learning machine with strong ability to deal with nonlinear problems is used to predict the end point of converter. The weight and threshold of LLM are optimized by genetic algorithm (GA) for the problem that random determination of weights and thresholds leads to instability of network structure. The effects of the hidden layer excitation function and the number of hidden layer nodes on the prediction results are compared. The simulation error data and prediction figure show that the improved method is effective. In order to improve the searching ability of PSO, the prediction accuracy is improved. In order to improve the searching ability of PSO, particle swarm optimization (PSO) algorithm is used to improve the searching ability of PSO. The inertia weight is improved and the influence of different functions and formulas on the optimization results is compared. The simulation results verify the effectiveness of the strategy. Finally, the software platform is built with VC, and the batching calculation model, prediction model and optimal control model are compiled into practical operation interface. The prediction and control of steelmaking process are realized.
【学位授予单位】：天津理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TF713;TP8

【相似文献】