摘要
板带钢是主要的钢材产品之一，广泛应用于汽车制造、食品包装等行业，对国民经济发展有着举足轻重的影响。随着科技的进步，用户对板带钢质量的要求越来越高。板形是带钢的重要质量指标，也是轧制领域研究的热点。近年来，人工智能方法以其在建模和控制方面的优势，在工业过程研究中得到了广泛的应用。本文选择基于改进最小二乘支持向量回归机(LS-SVR)的冷带轧机板形智能控制为研究课题，分析了现有板形智能控制方法的优缺点，重点研究了LS-SVR算法的改进，并利用它对板形控制系统进行了深入的研究。
首先，针对标准LS-SVR算法只适用于单输出系统回归估计，而工业过程多为多输出系统的问题，提出了一种多输出最小二乘支持向量回归机(MLSSVR)新算法，该算法秉承结构风险最小化原则，具有良好的泛化性能；针对MLSSVR超参数难以确定的问题，提出了一种基于粒子群算法的超参数优化方法，该方法不仅运算速度快，而且搜索能力强。
板形模式识别是板形控制的关键。考虑到板形精确控制的要求，采用一次、二次、三次和四次勒让德正交多项式表示板形的基本模式；并采用MLSSVR方法建立识别模型。研究结果表明，该方法可以有效确定板形缺陷的类型和含量，识别精度高，泛化能力强。
板形预测模型是板形控制系统设计的重要基础。为提高板形预测模型的准确性，以生产实测数据为基础，建立了基于MLSSVR的板形预测模型。仿真实验表明，MLSSVR板形预测模型具有预测精度高，鲁棒性强的优点。
最后，综合分析了影响矩阵控制方法和预测控制方法的特点，取长补短，提出了一种板形影响矩阵-预测控制方法。将该方法应用于某900HC可逆冷轧机进行仿真实验，结果表明，该方法较影响矩阵控制方法有更好的控制效果，是一种有效的板形控制方法。
Plate and strip steel, which is the main composition of the steel products and widelyused in automotive, food packaging industry, has significant influence to nationaleconomy. With the progress of science and technology, higher requirement to steel qualitywas made by customers. Flatness is one of the most important quality indexes of strip steel,and flatness controlling technique is the hot topic in the rolling area. In recent years,artificial intelligence has been used widely in industrial process study for its merits inmodeling, optimization and control. This paper choose the cold strip mill flatnessintelligent control based on improved Least Squares Support Vector Regression(LS-SVR)algorithms as research object. On analyzing the merits and defects of the existingintelligent control approach, improved LS-SVR algorithms was studied andcomprehensive research on flatness controlling system was made.
First of all, a novel Multi-output Least Squares Support Vector Regression (MLSSVR)approach was proposed to overcome the defects that standard LS-SVR algorithm, whichapplies only to more inputs single output system, can not use on more inputs more outputsindustrial process directly. The MLSSVR algorithm still meets the principle of StructuralRisk Minimization, therefore, keeps good generalization performance. Furthermore, totackle the difficulty in determining the hyper-parameters of MLSSVR, an optimizationmethod based on particle swarm optimization algorithm was adopted. This approach cannot only compute effectively, but also has strong searching ability.
Flatness pattern recognition is the key constituent of the flatness control system. Inorder to adapt to the higher demand of flatness controlling, flatness basic patternsexpressed by the linear, quadratic, cubic and quartic Legendre orthogonal polynomial wereproposed. And a novel flatness pattern recognition method based on MLSSVR was putforward. The results of experiment demonstrate that the proposed approach can distinguishthe types and define the magnitudes of the flatness defects effectively with high accuracy,high speed and strong generalization ability.
Flatness predictive model is the most important foundation of control system. Inorder to have higher precise flatness predictive model, a MLSSVR flatness predictive model is designed on the basis of measured data in production. Simulation experimentdemonstrates that the MLSSVR flatness predictive model has higher predictive accuracyand strong robustness.
Finally, effective matrix--predictive control approach was put forward oncomprehensively analyzing the characteristics of effective matrix control method andpredictive control method and combining the merits of the tow methods. Then, simulationexperiment on testing the performance of the control model was conducted on900HCreversible cold roll. It demonstrates that effective matrix--predictive control approach hasbetter control effect than effective matrix control method, therefore, is an effective flatnesscontrol method.
引文
[1] Liu Hong-min, He Hai-tao, Shan Xiu-ying. Flatness Control Based on Dynamic Effective Matrixfor Cold Strip Mills[J]. Chinese Journal of Mechanical Engineering,2009,22(2):287-296.
[2]翟博,孙荣生,孙建林.板形控制技术在鞍钢冷连轧机组的应用[J].鞍钢技术,2009,(1):49-52.
[3]张静,秦久莲.热轧带钢中板形的计算和控制[J].控制工程,2008,15(增刊):21-22,93.
[4]刘宏民,贾春玉.智能方法在板形控制中的应用[J].燕山大学学报,2010,34(1):1-5.
[5] John V. Ringwood, Senior Member. Shape Control Systems for Sendzimir Steel Mills[J]. IEEETransactions on Control Systems Technology, IEEE,2000,8(1):70-86.
[6] Shylu John, Sudipta Sikdar, P.Kumar Swamy. Hybrid Neural-GA Model to Predict and MinimizeFlatness Value of Hot Rolled Strips[J]. Journal of Materials Processing Technology,2008,5(195):314-32.
[7]张雪伟,王焱.智能识别方法在板形识别中的应用及发展趋势[J].钢铁研究学报,2010,22(1):1-3.
[8]华建新,周泽雁.冷轧带钢板形缺陷的多项式回归数学模型[J].钢铁,1992,27(3):27-31.
[9]刘进.冷轧带钢板形缺陷表达式回归及数学模型[J].轧钢,1996,(5):5-9.
[10]邸洪双,张晓峰,刘相华,等.冷轧薄带板形检测信号正交多项式分解及数学模型[J].钢铁,1995,30(9):33-36.
[11]周旭东,王国栋.冷轧板形正交多项式分解模型[J].钢铁,1997,32(8):46-47.
[12]戴江波,吴文彬.宝钢2030冷轧带钢板形识别和统计系统[J].北京科技大学学报,2003,25(6):572-576.
[13]刘建,王益群,孙福,等.基于粒子群理论的板形模糊模式识别方法[J].机械工程学报,2008,44(1):173-178.
[14]黄敏,董威,徐林,等.基于小波分析和神经网络的板形模式识别方法[J].模式识别与人工智能,2005,18(1):103-106.
[15]张秀玲,陈丽杰.基于径向基函数神经网络的板形模式识别研究[J].工业仪表与自动化装置,2009,(3):7-9.
[16]张秀玲,逄宗鹏.基于自适应神经模糊推理系统的板形模式识别[J].钢铁研究学报,2009,21(9):59-62.
[17]任海鹏,刘丁,郑岗.一种基于遗传算法的板形模式识别方法[J].重型机械,2002,(3):9-12.
[18]张秀玲,刘宏民.板形模式识别的GA-BP模型和改进的最小二乘法[J].钢铁,2003,38(10):29.
[19] Jia Chun-yu, Shan Xiu-ying, Liu Hong-min, et al. Fuzzy Neural Model for Flatness PatternRecognition[J]. Journal of Iron and Steel Research, International,2008,15(6):33-38.
[20]刘建昌,陈莹莹,张瑞友.基于POS-BP网络的板形智能控制器[J].控制理论与应用,2007,24(4):674-678.
[21]许东杰,贾春玉.基于量子粒子群算法的BP网络板形模式识别研究[J].燕山大学学报,2011,35(1):35-39.
[22]郑德忠,闫涛,王志勇.基于改进的微粒群算法的板形模式识别方法[J].冶金自动化,2007,(6):16.
[23]郑德忠,王志勇,闫涛.基于混沌优化的板形信号模式识别的研究[J].计量学报,2008,28(4):375.
[24]何海涛,李楠.基于SVM的改进RBF网络板形模式识别方法[J].自动化仪表,2007,28(5):1-4.
[25]张雪伟.基于多类支持向量机的板形模式识别方法[J].重型机械,2009,(3):7-11.
[26]赵丽娟,高丹,周宇.神经网络与有限元结合在轧机板形预报中的应用研究[J].重型机械,2007(3):5-8.
[27]黄亚飞.神经网络在热轧板凸度预报模型中的应用[J].山东冶金,2010,32(6):57-59.
[28]贾春玉.高精度宽带钢冷轧机板形模糊神经控制的研究[D].秦皇岛:燕山大学工学博士学位论文,2006:72-99,101-134.
[29]贾春玉,单修迎,牛召平.自调整动态神经网络模型及其在带材板形预测中的应用[J].钢铁研究学报,2006,18(12):50-53.
[30]许东杰.冷轧带钢平直度智能识别与预报模型研究[D].秦皇岛:燕山大学工学硕士学位论文,2011:43-65.
[31]陈杨.基于支持向量机预报模型的CVC轧机板形智能控制系统[J].机械设计与制造,2008,(11):120-122.
[32] Lu Bai-quan, Li Tian-duo, Lu Chong-de, et al. Wavelet Neural Network for Function Learning[J].Acta Automatic Sinica,1998,24(4):548-551.
[33] Khanmohammadi S, Hassanzadeh I, Sharifian M B B. Modified Adaptive Discrete ControlSystem Containing Neural Estimator and Neural Controller[J]. Artificial Intelligence inEngineering,2000,14(1):31-38.
[34] Gürocak H B. Genetic-algorithm-based Method for Tuning Fuzzy Logic Controllers[J]. FuzzySets and Systems,1999,108(1):39-47.
[35] Knohl T, Unbehauen H. Adaptive Position Control of Electrol Hydraulic Servo Systems UsingANN[J]. Mechatronics,2000,10(1-2):127-143.
[36] Woo Zhi-Wei, Chung Hung-Yuan, Lin Jin-Jye. PID Type Fuzzy Controller with Self-tuningScaling Factors[J]. Fuzzy Sets and Systems,2000,115(2):321-326.
[37] Dumitrache Ion, Buiu Catalin. Genetic Learning of Fuzzy Controller[J]. Mathematics andComputers in Simulation,1999,49(1-2):13-26.
[38] J. Y. Jung. Development of Fuzzy Control Algorithm for Shape Control in Cold Rolling[J].Journal of Materials Processing Technology,1995,48:187-195.
[39] J. Y. Jung. Fuzzy-control Simulation of Cross-sectional Shape in Sex-high Cold-rolling Mills[J].Journal of Materials Processing Technology,1996,61(1-3):61-69.
[40] J. Y. Jung. Simulation of Fuzzy Shape Control for Cold-rolled Strip with Randomly Irregular StripShape[J]. Journal of Materials Processing Technology,1997,63(1-3):248-235.
[41] Morooka Y. Shap Control of Rolling Mills by a Neural and Fuzzy Hybrid Architecture[J].International Joint Conference of the Fourth IEEE International Conference on Fuzzy Systems andthe Second International Fuzzy Engineering Symposium, Proceedings of1995IEEE InternationalConference,1995,5(20-24):47-48.
[42]张秀玲.冷带轧机板形智能识别与智能控制研究[D].秦皇岛:燕山大学工学博士学位论文,2002:20-45,77-99.
[43]何海涛.宽带钢冷轧机板形在线控制智能模型的研究与应用[D].秦皇岛:燕山大学工学博士学位论文,2008:41-75.
[44]张秀玲,刘宏民.板形控制的传递矩阵方法[J].机械工程学报,2003,39(11)100-103.
[45]张秀玲,逄宗鹏. ANFIS的板形控制动态影响矩阵方法[J].智能系统学报,2010,5(4):360-365.
[46]贾春玉,崔艳超,许东杰.基于非线性预测模型的单神经元自适应PID板形控制[J].冶金设备,2010(6):1-5.
[47]何海涛,张兰.一种基于聚类的板形控制模糊神经网络模型[J].重型机械,2008,(2):5-9.
[48]周晓敏,张清东,王长松.基于Hopfield神经网络板形预测控制模型[J].上海金属,2007,29(2):44-47.
[49]王京,李洪全.基于混合粒子群算法的液压弯辊控制[J].机床与液压,2007,35(10):152-154.
[50]单修迎.四辊冷轧机板形神经模糊控制研究[D].秦皇岛:燕山大学工学硕士学位论文,2007:71-93.
[51]李丽娟.最小二乘支持向量机建模及预测控制算法研究[D].杭州:浙江大学工学博士学位论文,2008:3,4,80-81.
[52]张学工.统计学习理论本质[M].北京:清华大学出版社,2000:168-236.
[53]邓乃扬,田英杰.数据挖掘中的新方法—支持向量机[M].北京:科学出版社,2004:49-50,132-188,228-245,356-358.
[54]边肇祺,张学工.模式识别(第2版)[M].北京:清华大学出版社,2000:122-205.
[55]柳回春,马树元.支持向量机的研究现状[J].中国图像图形学报,2007,7(6):618-623.
[56]杨滨.智能计算与应用研究[D].长春:吉林大学工学博士学位论文,2010:1.
[57] Boser B, Guyon I, Vapnik V. A Training Algorithm for Optimal Margin Classifiers[C].Proceedings of the Fifth Annual ACM Workshop on Computational Learning Theory. New York:ACM Press,1992:144-152
[58] Osuna E, Freundr. An Improved Training Algorithm for Support Vector Machines[C]. IEEEWorkshop on Neural Networks and Signal Processing, Amelia Island,1997:276-285.
[59] Joachimst. Making Large-scale Support Vector Machine Practical[C]. Advances in KernelMethods Support Vector Learning, Cambridge, Massachusetts: The MIT Press,1999:169-184.
[60] J. C. Platt. Fast Training of Support Vector Machines Using Sequential Minimal Optimization[C].Advances in Kernel Methods-Support Vector Learning, Cambridge, Massachusetts: The MITPress,1999:185-208.
[61] Cauwenberghs G, Poggio T. Incremental and Decremental. Support Vector Machine Learning[C].Advances in Neural Information Processing Systems(NIPS2000). Cambridge, MA: MITPress,2001:108-122.
[62]李忠伟,张健沛,杨静.基于支持向量机的增量学习算法研究[J].哈尔滨工程大学学报,2005,26(5):643-646.
[63]顾亚祥,丁世飞.支持向量机研究进展[J].计算机科学,2011,38(2):14-17.
[64] Tang Yu-chun, Jin Bo, Zhang Yan-qing. Granular Support Vector Machines with AssociationRules Mining for Protein Homology Prediction[J]. Artificial Intelligence in Medicine,2005,35(1):121-134.
[65]张鑫,王文剑.一种基于粒度的支持向量机学习策略[J].计算机科学,2008,35(8A):101-103,116.
[66]文贵华,向君,丁月华.基于商空间粒度理论的大规模SVM分类算法[J].计算机应用研究,2008,25(8):2299-2301.
[67] Suykens J A K, Vandewalle J. Least Squares Support Vector Machine Classifiers[J]. NeuralProcessing Letters,1999,9(3):293-299.
[68] Gestel T V, Suykens J A K. Benchmarking Least Squares Support Vector Machine Classifiers[J].Machine Learning,2004,54(1):5-32.
[69] Anguital D, Boni. A Digital Least Squares Support Vector Machines[J]. Neural Processing Letters,2003,18(1):65-72.
[70] Tsujinishi D, Abe S. Fuzzy Least Squares Support Vector Machines for Multiclass Problems[J].Neural Networks,2003,16(5,6):785-792.
[71]范玉刚,李平,宋执环.动态加权最小二乘支持向量机[J].控制与决策,2006,21(10):1129-1134.
[72]吴春国.广义染色体遗传算法与迭代式最小二乘支持向量机回归算法研究[D].长春:吉林大学工学博士学位论文,2006:49-57.
[73]叶美盈,汪晓东,张浩然.基于在线最小二乘支持向量机回归的混沌时间序列预测[J].物理学报,2005,54(6):2568-2573.
[74]张浩然,汪晓东.回归最小二乘支持向量机的增量和在线式学习算法[J].计算机学报,2006,29(3):400-406
[75] Suykens J A K, Lukas L. Sparse Approximation Using Least Squares Support VectorMachines[C]. Proc. of the IEEE International Symposium on Circuits and Systems (ISCAS2000),Geneva, Switzerland,2000:757-760.
[76]赵永平,孙健国.一种快速稀疏最小二乘支持向量回归机[J].控制与决策,2008,23:1347-1352.
[77] Scholkopf B, Smola A, Williamson C, etal. New Support Vector Algorithms[J]. NeuralComputation,2000,12(5):1207-1245.
[78] Chin W H. A Simple Decomposition Method for Support Vector Machines[R].Technical reportNational Taiwan University,1999:52-59.
[79] Chun Fu Lin, Sheng De Wang. Fuzzy Support Vector Machines[J]. IEEE Transactions on NeuralNetworks,2002,13(2):464-471.
[80] Yang M H, A Huja N. A Geometric Approach to Train Support Vector Machines[C]. Proceedingsof CVPR2000, Hilton Head Island,2000:430-437.
[81]张文生,丁辉,王钰.基于邻域原理计算海量数据支持向量的研究[J].软件学报,2001,12(5):712-720.
[82]汪西莉,焦李成.一种基于马氏距离的支持向量快速提取算法[J].西安电子科技大学学报(自然科学版),2004,31(4):639-643.
[83]朱杰,吴树芳.支持向量机研究现状[J].大众科技,2009,(5):88-89.
[84] Knerr S. Single-layer Learning Revisited: A Step Wise Procedure for Building and Training aNeural Network[C]. Neurocomputing: Algorithms, Architectures and Applications, NATO ASLSpringer,1990:35-42.
[85] Platt J, Cristianini N, Shawe Taylor J. Large Margin DAGs for Multiclass Classification[C].Advances in Neural Information Processing Systems, Massachusetts: MIT Press,2000:547-553.
[86]宋召青,崔和.支持向量机理论的研究与进展[J].海军航空工程学院学报,2008,23(2):143-152.
[87]张学工.关于统计学习理论与支持向量机[J].自动化学报,2000,26(1):32-41.
[88] J.A.K. Suykens, J. Vandewalle. Least Squares Support Vector Machine Classifiers[J]. NeuralProcessing Letter,1999,9(3):293-300.
[89]胡昌华,蔡艳宁.基于多重回归LSSVM的并发故障诊断[J].华中科技大学学报(自然科学版),2009,37(I):1-5.
[90] Gestel T V, J.A.K. Suykens. Benchmarking Least Squares Support Vector Machine Classifiers[J].Machine Learning,2004,54(1):5-32.
[91]王兴玲,李占斌.基于网格搜索的支持向量机核函数参数的确定[J].中国海洋大学学报,2005,35(5):859-862.
[92] Mathias M. Adankon, Mohamed Cheriet. Model Selection for the LS-SVM. Application toHandwriting Recognition[J]. Pattern Recognition,2009(42)3264-3270.
[93]郭新辰.最小二乘支持向量机算法及应用研究[D].长春:吉林大学工学博士学位论文,2008:6-7,9-10,21-35
[94]陶少辉.最小二乘支持向量机的改进及其在化学化工中的应用[D].杭州:浙江大学工学博士学位论文,2006:9,11.
[95] Vapnik V N. The Nature of Statistical Learning Theory[M]. NY: Springer-Verlag,1995.
[96] Burges C J C. A Tutorial on Support Vector Machines for Pattern Recognition[J]. Data Mining andKnowledge Discovery,1998,2(2).
[97]李建民,张拔,林福宗.支持向量机的训练算法[J].清华大学学报(自然科学版),2003,43(l):120-124.
[98]孙丰阔.基于支持向量回归(SVR)的线性相位FIR滤波器设计[D].厦门:厦门大学工学硕士学位论文,2009:26.
[99]余艳芳.改进型支持向量回归机及其在过程建模与控制中的应用[D].上海:华东理工大学工学博士学位论文,2010:38.
[100] Suykens J A K, Van Gestel T, De Brabanter J, et al. Least Squares Support Vector Machines[M].Singapore: World Scientific Publishing Co Pte Lte,2002.
[101] An Sen-jian, Liu Wan-quan, Svetha Venkatesh. Fast Cross-validation Algorithms for LeastSquares Support Vector Machine and Kernel Ridge Regression[J]. Pattern Recognition,2007,40(1):2154-2162.
[102]单修迎,刘宏民.含有三次板形的新型板形模式识别方法[J].钢铁,2010,45(8):56-60.
[103]伊国芳,王益群,孙旭光.基于神经网络的板形信号模式识别方法的研究[J].中国机械工程,2004,15(24):15-18.
[104]张秀玲,陈丽杰. RBF神经网络的板形预测控制[J].智能系统学报,2010,5(1):70-73.
[105] Burnhamr, Colei, Gentilea, et al. Model Based Flatness Control of Thin Strip and Foil[J]. Ironand Steel,2003,38(6):36-40.
[106]贾春玉,王建国,李兴东,等.人工智能技术在板形控制中的应用[J].冶金设备,2003,8(4):8-13.
[107]刘建昌,王柱.基于神经网络模式识别的板形模糊控制器[J].东北大学学报(自然科学版),2005,26(8):718-721.
[108]舒迪前.预测控制系统及其应用[M].北京:机械工业出版社,1996:66-67.
[109]包哲静.支持向量机在智能建模和模型预测控制中的应用[D].杭州:浙江大学工学博士学位论文,2007:4-12.
[110]焦巍,刘光斌.非线性模型预测控制的智能算法综述[J].系统仿真学报,2008,20(24):6581-6586
[111]苏成利.非线性模型预测控制的若干问题研究[D].杭州:浙江大学工学博士学位论文,2006:4.
[112] Zhong Wei-min, Pi Dao-ying, Sun You-xian. Study on SVM Based Model Predictive Control[C].Proceedings of the Fifth Word Congress on Intelligent Control and Automation(in Chinese),2004, Hangzhou, China.USA: IEEE,2004:607-610.
[113] Zhong Wei-min, Pi Dao-ying, Sun You-xian. An Approach of Nonlinear Model Multi-step-aheadPredictive Control Based on SVM[J]. Lecture Notes in Computer Science(S0302-9743),2005,3516(1):1036-1039.
[114] Zhong Wei-min, He Guo-long, Pi Dao-ying, et al. SVM with Quadratic Polynomial KernelFunction Based Nonlinear Model One-step-ahead Predictive Control[J]. Chinese J.Chem.Eng(S1004-9541),2005,13(3):373-379.
[115] Man Gyun Na, Belle R Upadhyaya. Model Predictive Control of an SP-100Space ReactorUsing Support Vector Regression and Genetic Optimization[J]. IEEE Transactions on NuclearScience(S0018-9499),2006,53(4):2318-2327.
[116] Xi Xue-Cheng, Aun-Neow Poo, Siaw-Kiang Chou. Support Vector Regression Model PredictiveControl on a HVAC plant[J]. Control Engineering Practice(S0967-0661),2007,15(8):897-908.
[117]王宇红,黄德先,高东杰,等.基于LS-SVM的非线性预测控制技术[J].控制与决策,2004,19(4):383-387.
[118]王宇红,黄德先,高东杰,等.基于支持向量机的非线性预测控制技术[J].信息与控制,2004,33(2):133-136.
[119]张浩然,韩正之,李昌刚.基于支持向量机的非线性模型预测控制[J].系统工程与电子技术,2003,25(3):330-334.