锻压技术

基于机器学习的热轧轧制力预测

英文标题：Prediction of rolling force in hot rolling based on machine learning

作者：刘明华张强刘英华王文礼

单位：西安建筑科技大学

分类号：TF16

出版年,卷(期):页码：2021,46(10):233-241

摘要：

针对传统轧制力数学模型预测精度差的问题，基于板带轧制数据和支持向量回归（SVR）模型对轧制力进行预测。采用主成分分析（PCA）技术来降低输入变量的维数，同时提出了改进粒子群优化（IPSO）算法调节惯性权值和加速因子，并采用IPSO算法对SVR模型中的惩罚因子c、核函数参数g和不敏感损失参数ε进行优化，最终建立PCA-IPSO-SVR轧制力预测模型。与PCA-PSO-SVR、PSO-SVR和Grid-SVR模型相比，PCA-IPSO-SVR模型的3种误差指标处于最低水平，且平均绝对百分比误差（MAPE）为4.8153%。仿真结果表明：与常规PSO算法相比，IPSO算法可以避免陷入局部极小值，从而获得模型最优参数和提高模型预测精度；与其他3种模型相比，PCA-IPSO-SVR模型具有较高的预测精度和较好的泛化性能。

In view of the poor prediction accuracy for the traditional rolling force mathematical model, the rolling force was predicted based on the strip rolling data and the support vector regression (SVR) model. Principal component analysis (PCA) technology was employed to reduce the dimension of input variables, and an improved particle swarm optimization (IPSO) algorithm was proposed to regulate the inertia weight and acceleration factors. The penalty factor c , kernel function parameter g and insensitive loss parameter ε of the SVR model were optimized by the IPSO algorithm. Finally, PCA-IPSO-SVR rolling force prediction model was established. Compared with the PCA-PSO-SVR, PSO-SVR and Grid-SVR models, the three error indexes of the PCA-IPSO-SVR model were at the lowest level, and the average absolute percentage error (MAPE) value was 4.8153%. The simulation results show that compared with the conventional PSO algorithm, the IPSO algorithm can avoid falling into the local minimums, thereby obtaining the optimal parameters of the model and improving the prediction accuracy of the model. Compared with the other three models, the PCA-IPSO-SVR model has higher prediction accuracy and better generalization performance.

基金项目：

作者简介：刘明华(1976-），男，博士，副教授 E-mail：lmhxauat@163.com 通信作者：王文礼(1977-），男，博士，教授 E-mail：wangwl@nwpu.edu.cn

参考文献：

[1]Bagheripoor M, Bisadi H. Application of artificial neural networks for the prediction of roll force and roll torque in hot strip rolling process[J]. Applied Mathematical Modelling, 2013, 37(7): 4593-4607.

[2]Wang Z H, Zhang D H, Gong D Y, et al. A new data-driven roll force and roll torque model based on FEM and hybrid PSO-ELM for hot strip rolling[J]. ISIJ International, 2019, 59(9): 1604-1613.

[3]刘相华, 赵启林, 黄贞益. 人工智能在轧制领域中的应用进展[J]. 轧钢, 2017, 34(4): 1-5.

Liu X H, Zhao Q L, Huang Z Y. Prospect of artificial intelligent application in rolling[J]. Steel Rolling, 2017, 34(4): 1-5.

[4]Chun M S, Biglou J, Lenard J G, et al. Using neural networks to predict parameters in the hot working of aluminum alloys[J]. Journal of Materials Processing Technology, 1999, 86: 245-251.

[5]马威, 李维刚, 赵云涛, 等. 基于深度学习的热连轧轧制力预测[J]. 钢铁研究学报, 2019, 31(9): 805-815.

Ma W, Li W G, Zhao Y T, et al. Prediction of hot-rolled roll force based on deep learning[J]. Journal of Iron and Steel Research, 2019, 31(9): 805-815.

[6]周富强, 曹建国, 张杰, 等. 基于神经网络的冷连轧机轧制力预报模型[J]. 中南大学学报：自然科学版, 2006, 37(6): 1155-1160.

Zhou F Q, Cao J G, Zhang J, et al. Prediction model rolling force for tandem cold rolling mill based on neural networks and mathematical models[J]. Journal of Central South University：Science and Technology, 2006, 37(6): 1155-1160.

[7]王前锋. 基于改进型支持向量机算法的轧机轧制力预测[J]. 锻压技术, 2019, 44(4): 131-137.

Wang Q F. Rolling force prediction of rolling mill based on improved support vector machine algorithm[J]. Forging & Stamping Technology, 2019, 44(4): 131-137.

[8]何飞, 石露露, 黎敏, 等. 基于多模态和加权支持向量机的热轧轧制力智能预报[J]. 工程科学学报, 2015, 37(4): 517-521.

He F, Shi L L, Li M, et al. Intelligent prediction of rolling force in hot rolling based on a multi-model and weighted support vector machine[J]. Chinese Journal of Engineering, 2015, 37(4): 517-521.

[9]Guo Z Y, Sun J N, Du F S. Application of finite element method and artificial neural networks to predict the rolling force in hot rolling of Mg alloy plates[J]. Journal of the Southern African Institute of Mining and Metallurgy, 2016, 116(1): 43-48.

[10]柏阳, 吴玉程, 罗志勇, 等. 基于Arrhenius方程和BP神经网络的2024Al/Al18B4O33w复合材料热变形流变应力预测[J]. 锻压技术, 2019, 44(8): 168-175.

Bo Y, Wu Y C, Luo Z Y, et al. Prediction on hot deformation flow stress of 2024Al /Al18B4O33w composites based on Arrhenius equation and BP neural network[J]. Forging & Stamping Technology, 2019, 44(8): 168-175.

[11]张生, 姜万录, 张佳慧. 基于支持向量机预测的冷连轧机轧制力精确设定方法研究[J]. 液压与气动, 2017,41(7): 50-55.

Zhang S, Jiang W L, Zhang J H. SVM prediction-based rolling force setting calculation method of tandem cold rolling mill[J]. Chinese Hydraulics & Pneumatics, 2017, 41(7): 50-55.

[12]魏立新, 魏新宇, 孙浩, 等. 基于改进遗传算法优化SVM的轧制力预报[A]. 第37届中国控制会议论文集[C]. 皮斯卡塔韦新泽西州: IEEE, 2018.

Wei L X, Wei X Y, Sun H, et al. Rolling force prediction of SVM based on improved genetic algorithm[A]. Proceedings of the 37th Chinese Control Conference[C]. Piscataway NJ: IEEE, 2018.

[13]Wu D S, Yang Q, Wang D Z. Rolling force prediction based on PSO optimized support vector regression[A]. 2011 Seventh International Conference on Natural Computation[C]. Piscataway NJ: IEEE, 2011.

[14]Wang D H, Tan D, Liu L. Particle swarm optimization algorithm: An overview[J]. Soft Computing, 2017, 22(2): 387-408.

[15]Peng Z, Jiang Y, Yang X, et al. Bus arrival time prediction based on PCA-GA-SVM[J]. Neural Network World, 2018, 28（1）: 87-104.

[16]Sims R B. The calculation of roll force and torque in hot rolling mills[J]. Proceedings of the Institution of Mechanical Engineers, 1954, 168(1): 191-200.

[17]Box J F. Guinness, gosset, fisher, and small samples[J]. Statistical Science, 1987, 2(1): 45-52.

[18]Vapnik V N. The Nature of Statistical Learning Theory[M]. New York: Spring, 1995.

[19]Eberhart R, Kennedy J. A new optimizer using particle swarm theory[A]. Proceedings of the 6th International Symposium on Micro Machine and Human Science[C]. IEEE Industrial Electronics Society: IEEE Service Center, 1995.

服务与反馈：

【文章下载】【加入收藏】