精轧温度模型优化算法与控制策略的研究  

Temperature Optimization and Control Strategy of Hot Rolled in Finishing

作　　者：张博[1,2] 郭强[1] 张飞[1] 宗胜悦[1] 张佳庆[1] 

机构地区：[1]北京科技大学高效轧制国家工程研究中心,北京100083 [2]中债资信评估有限责任公司,北京100031

出　　处：《控制工程》2014年第3期352-356,共5页Control Engineering of China

基　　金：新世纪优秀人才支持计划项目(NCET-10-0223)

摘　　要：针对国内热轧生产线上精轧区测温点稀疏、温度精度要求高的特点,采用二维有限差分方法建立了精轧区板坯温度的机理模型,通过现场实测数据并结合粒子群算法对温度模型进行优化。针对轧制过程中影响因素多、环境复杂的特点,将支持向量机引入精轧温度模型的预测,并与机理模型和实测数据对比,检测其有效性。经过验证,所建模型的计算精度与实测数据误差在10℃以内。利用所建模型,对终轧区喷水量和加速度提出一种综合控制策略,可为实际生产提供参考。Temperature is an important factor in rolling process. It can directly affect the mechanical properties and precision of the roll- ing force. At present, the domestic production lines use surface temperature as the prediction for the average rolling temperature. The forecast error caused by this means is not helpful to improve product quality. In this paper, the author used method of a two-dimensional finite difference to establish a model of the temperature in the finishing stands of hot rolled, and used a explicit difference method to solve the model. This model was optimized by collected data through particle swarm optimization. Considering that there are too many factors in the rolling process and the environmental complexity, this paper introduces support vector machine （SVM） into the temperature model prediction. The efficiency of SVM has been shown by comparing it with the mechanism model. The difference between this model and the measure data is less than ＋ 10℃. Based on the model, an integrated control strategy is given.

关 键 词：精轧 热连轧 温度模型 粒子群算法 支持向量机 

分 类 号：TP27[自动化与计算机技术—检测技术与自动化装置]