钢板大变形热轧机轧制力DEI-RBF预测研究  

Research on Rolling Force DEI-RBF Prediction of Large Deformation Hot Mill

作　　者：王保华[1] 葛新锋 杨波[3] 胡草笛 WANG Bao-hua;GE Xin-feng;YANG Bo;HU Cao-di(School of Mechanical and Electrical Engineering,Jiaozuo University,He’nan Jiaozuo 454000,China;Engineering Technology Center,Xuchang University,He’nan Xuchang 461000,China;School of Mechanical and Power Engineering,Henan Polytechnic University,He’nan Jiaozuo 454003,China)

机构地区：[1]焦作大学机电工程学院,河南焦作454000 [2]许昌学院工程技术中心,河南许昌461000 [3]河南理工大学机械与动力工程学院,河南焦作454003

出　　处：《机械设计与制造》2023年第9期217-220,共4页Machinery Design & Manufacture

基　　金：河南省高等学校优秀基层教学组织(教高[2017]730号);河南省科技厅科技攻关项目(182102210508)。

摘　　要：支持向量机实际计算过程的复杂性主要由支持向量数决定,可以获得优异鲁棒性,精度也获得明显提升。设计了一种通过差分进化改进支持向量机模型(DEI-RBF),分并以RBF核函数支持向量机(RBF-SVM)构建初始模型。通过差分进化算法完成RBF-SVM惩罚系数C以及RBF核函数参数σ的寻优,结果表明DEI-RBF可以实现热轧轧制力的精确预测,达到现场使用要求。研究结果表明:以RBF核函数构建的支持向量机回归模型获得了最大的R2,同时均方差(MSE)以及平均绝对误差(MAE)都达到了最小,显著提升了模型效果。采用差分进化算法进行优化后的支持向量机回归模型获得了更优性能,预测误差在5%以内的概率为99.2%,相对传统轧制力计算模型获得了更高预测准确性。The complexity of the actual calculation process of support vector machine is mainly determined by the number of sup⁃port vectors,which can obtain excellent robustness and improve the accuracy significantly.A differential evolution improved sup⁃port vector machine model(DEI-RBF)was designed,and the initial model was constructed by dividing and using RBF kernel support vector machine(RBF-SVM).The differential evolution algorithm was used to optimize the RBF-SVM penalty coefficient C and the kernel function parameterσof RBF.The results show that DEI-RBF can accurately predict the rolling force of hot strip rolling mill,so as to meet the requirements of field application.The results show that the support vector machine regression model constructed by the RBF kernel function achieves the maximum R2,while the mean square error(MSE)and mean absolute error(MAE)reach the minimum,which significantly improves the model effect.The SVM regression model optimized by differential evolution algorithm has better performance,and the probability of prediction error within 5%is 99.2%.Compared with the tradi⁃tional rolling force calculation model,the SVM regression model has higher prediction accuracy.

关 键 词：轧制力预测 核函数 支持向量机 差分进化 

分 类 号：TH16[机械工程—机械制造及自动化]