基于SSA一SVR模型的步进式加热炉炉温预测
2025-08-20牛佳丽刘丕亮崔桂梅任春妮
中图分类号:TB9;TF31 文献标志码:A文章编号:1674-5124(2025)07-0064-08
Abstract: The prediction and temperature control of heating furnace temperature is of great significance to improve the quality of billet, energy saving and consumption reduction. Aiming at the problems such as low accuracy of heating furnace temperature prediction, a furnace temperature prediction model (SSA-SVR) based on the combination of Sparrow Search Algorithm (SSA) and Support Vector Machine Regression (SVR) is proposed from the data-driven point of view.By comparing this prediction model with five other prediction models,the results show that the SSA-SVR model has the smallst mean square error (MSE) index and the highest goodness of fit (r2) , and the accuracy of the model is significantly improved compared with the SVR model, with the mean square error index significantly reduced and the goodness of fit improved by 0.0283. It provides a powerful support for the improvement of the temperature control accuracy of the furnace, and provides a powerful support for the improvement of the furnace temperature control accuracy of the steel furnace. Provide strong support for improving the control accuracy of the furnace temperature, and provide a more reliable basis for billet rolling.
Keywords: heating furnace roling; sparrow search optimization algorithm; support vector machine; furnace temperature prediction
0 引言
目前,我国正在积极推进\"双碳\"政策,该政策对我国环境质量改善和产业结构转型有着重要的指导作用,钢铁行业的绿色转型对我国环境改善以及钢铁行业的高质量发展有着重要意义[1]。数据表明:加热炉是钢铁工业中能耗占比较大的设备,能耗大约占整个行业的 25%[2] 。加热炉的作用是将钢坏从常温加热至适合工艺要求的轧制温度,加热炉的整体框架结构如图1所示。在生产中,钢铁的质量与其加工时的温度密切相关,因此建立加热炉炉温预报模型可以为钢坏提供可靠的温度环境,有效做到节能降耗,响应\"双碳\"政策。

目前,加热炉炉温建模主要包括有三种。第一种是根据已知的原理和定律分析整个过程的内部机理,称为机理建模。王中杰、柴天佑等人根据已知加热炉内钢坏的传热机理模型结合神经元网络成功辨识出钢坏表面的温度曲线,并在此基础上建立出新的钢坯温度预报的边界条件模型,得出自适应钢坏温度预报模型[3]。第二种是多元回归分析建模,主要为主元回归和最小二乘法。周其节采用一种由脉冲响应序列估计传递函数的最小二及传递函数乘法,辨识出系统的参数[4]。第三种是应用采集到的数据信息,利用神经网络进行数据建模[5]。王锡淮等人建立了基于模糊聚类的径向基神经网络预测模型,并根据该控制对象数学模型研发一套加热炉炉温控制系统。……
