基于机理和数据融合的热轧终轧温度集成建模  

Integrated modeling of finishing delivery temperature during hot rolling process based on data-mechanism fusion method

作　　者：宋君 武文腾 王奎越 孙杰[2] 彭文[2] 曹忠华 SONG Jun;WU Wenteng;WANG Kuiyue;SUN Jie;PENG Wen;CAO Zhonghua(Ansteel Beijing Research Institute Co.,Ltd.,Beijing 102211,China;State Key Laboratory of Rolling and Automation,Northeastern University,Shenyang 110819,China)

机构地区：[1]鞍钢集团北京研究院有限公司,北京102211 [2]东北大学轧制技术及连轧自动化国家重点实验室,辽宁沈阳110819

出　　处：《冶金自动化》2023年第5期55-62,共8页Metallurgical Industry Automation

基　　金：国家重点研发计划项目(2022YFB3304800)。

摘　　要：在板带热轧过程中,终轧温度的精准预测是产品三维尺寸和产品性能控制的关键。为了提升终轧温度的预测精度,提出了一种数据机理融合的预测建模方法。该方法在混合特征选择基础上,融合机理模型结果以改进数据集,并引入哈里斯鹰优化算法(Harris hawks optimization,HHO)优化预测模型,实现了终轧温度的高精度预测。计算结果表明,优化后的融合模型的平均绝对误差EMA(mean absolute error,MAE)、均方误差EMS(mean square error,MSE)和R2分别达到4.1368、31.97和0.9322,预测偏差在15℃以内的数据占比由94.33%提升至98.25%,能够实现高精度的热轧终轧温度预测。During the hot rolling process,the accurate prediction of finishing delivery temperature is crucial for controlling the three-dimensional dimensions and performance of the product.In order to improve the prediction accuracy of finishing delivery temperature,a data-mechanism fusion prediction modeling method was proposed.This method improves the data quality by fusing the results of the mechanism model into the data set based on hybrid feature selection,and introduces the Harris hawks optimization algorithm to optimize the prediction model,achieving high-precision prediction of finishing delivery temperature.The calculated results show that the optimized fusion model achieves EMA(mean absolute error,MAE),Ems(mean square error,MSE),and R?of 4.1368,31.97,and 0.9322,respectively,and the proportion of data with prediction deviation within 15℃is improved from 94.33%to 98.25%.Therefore,the proposed method can achieve high-precision prediction of the finishing delivery temperature in hot rolling.

关 键 词：热连轧 终轧温度 融合模型 集成建模 哈里斯鹰优化算法 

分 类 号：TG335[金属学及工艺—金属压力加工]