转炉提钒静态模型研究
|
摘要
钒是一种极为重要的金属，广泛用于冶金、有色金属加工、化工、电子和航空航天等方面。我国有丰富的钒钛磁铁资源，是世界四大产钒国之一。转炉提钒是我国主要的生产方式。转炉提钒是一个多元多相的高温化学反应过程，反应机理复杂，影响因素众多，受原材料、工艺环境变化的影响较大，不便于人工控制。为提高提钒技术水平，改变生产过程中存在的指标波动大的缺点，需要对生产过程实现计算机控制，以减少提钒操作中人为因素的影响。
论文从工程应用的角度出发，针对我国转炉提钒手工操作的现状，在开发提钒静态模型的过程中，以转炉提钒为特定研究对象，考虑了传统数学模型适应性差、难于移植的缺点，利用RBF神经网络和遗传算法等人工智能技术开发出具有较好自学习和自适应能力的静态模型。
论文首先介绍了建立提钒静态模型的意义，讨论了国内外转炉模型的研究应用情况及发展趋势，深入分析了各种建模方法的优缺点。针对提钒过程的复杂性，提出转炉提钒建模应以神经网络为主要技术手段，结合其他人工智能技术，建立起比传统数学模型具有更高精度的模型。
论文结合工艺原理和实际提钒过程的特点，找出提钒过程的内在影响因素和影响静态模型的主要因素。然后分别应用RBF神经网络建立吹氧时间子模型、冷却剂子模型，并结合遗传算法完成RBF神经网络本身其神经元个数、Spread等参数的优化，使模型具有较强的泛化能力。仿真结果表明该算法具有较快的收敛速度和较强的学习能力。
最后，本文使用面向对象的可视化Visual C++ 6 .0编程语言，开发出提钒静态控制模型软件。
Vanadium, which is application to metallurgy, chemistry, electronic and air mechanic etc widely, is an important metal. Our country is abundant in natural resources containing Vanadium, Titanium, etc. Now our country has become one of the four greatest produced devanadium countries in the world. Oxygen converter devanadium now has become the main production method in our country. The process of devanadium in which complex mechanism of reaction and many affect factors is a high temperature chemical reaction process, influenced by the variety of raw material and technics condition, so it is very inconvenient to operate manually. In order to improve the level of devanadium technology and correct the defect of fluctuation of the target in production process, it is necessary for us to realise the computer control in production process and reduce the influence by the manual factor.
On the process of developing the static devanadium model, thinking of the status that the devanadium process is still manual operation in our country, the oxygen converter devanadium is chosen as a specific research object from the viewpoint of industrial application in this paper. And on the basis of considering the demerits of the traditional mathematical model, the paper has developed a perfect self-study and self-adaptive static model by means of some artificial intelligence technology such as RBF networks and Genetic Algorithms etc.
Firstly this paper introduce the meaning of developing devanadium static model and discuss the application and development of the domestic and international research, and then thoroughly analyze the merits and demerits in these model-making methods. Aiming at the complexity of devanadium process, this paper put forward an advice that the more precise model ,then traditional mathematical model should be made by means of the NN and other AI technology.
On the basis of combining the characteristic of the technics theory and practical devanadium process, this paper conclude the internal affect factor of devanadium process and the main factor of static model. Then the author has constructed the submodel of the time of blowing oxygen and refrigerant, making use of the RBF neural networks. Meantime, applying the genetic algorithm, completed the optimizationof parameter of the number of neuron and spread, and get the model with a better generalization ability. The imitate results show that the algorithm has an ability of fast
convergence speed and strong study power.
At last, this paper develop a software for devanadium static control model by means of object-oriented programming language Visual C++ 6.0.
论文从工程应用的角度出发，针对我国转炉提钒手工操作的现状，在开发提钒静态模型的过程中，以转炉提钒为特定研究对象，考虑了传统数学模型适应性差、难于移植的缺点，利用RBF神经网络和遗传算法等人工智能技术开发出具有较好自学习和自适应能力的静态模型。
论文首先介绍了建立提钒静态模型的意义，讨论了国内外转炉模型的研究应用情况及发展趋势，深入分析了各种建模方法的优缺点。针对提钒过程的复杂性，提出转炉提钒建模应以神经网络为主要技术手段，结合其他人工智能技术，建立起比传统数学模型具有更高精度的模型。
论文结合工艺原理和实际提钒过程的特点，找出提钒过程的内在影响因素和影响静态模型的主要因素。然后分别应用RBF神经网络建立吹氧时间子模型、冷却剂子模型，并结合遗传算法完成RBF神经网络本身其神经元个数、Spread等参数的优化，使模型具有较强的泛化能力。仿真结果表明该算法具有较快的收敛速度和较强的学习能力。
最后，本文使用面向对象的可视化Visual C++ 6 .0编程语言，开发出提钒静态控制模型软件。
Vanadium, which is application to metallurgy, chemistry, electronic and air mechanic etc widely, is an important metal. Our country is abundant in natural resources containing Vanadium, Titanium, etc. Now our country has become one of the four greatest produced devanadium countries in the world. Oxygen converter devanadium now has become the main production method in our country. The process of devanadium in which complex mechanism of reaction and many affect factors is a high temperature chemical reaction process, influenced by the variety of raw material and technics condition, so it is very inconvenient to operate manually. In order to improve the level of devanadium technology and correct the defect of fluctuation of the target in production process, it is necessary for us to realise the computer control in production process and reduce the influence by the manual factor.
On the process of developing the static devanadium model, thinking of the status that the devanadium process is still manual operation in our country, the oxygen converter devanadium is chosen as a specific research object from the viewpoint of industrial application in this paper. And on the basis of considering the demerits of the traditional mathematical model, the paper has developed a perfect self-study and self-adaptive static model by means of some artificial intelligence technology such as RBF networks and Genetic Algorithms etc.
Firstly this paper introduce the meaning of developing devanadium static model and discuss the application and development of the domestic and international research, and then thoroughly analyze the merits and demerits in these model-making methods. Aiming at the complexity of devanadium process, this paper put forward an advice that the more precise model ,then traditional mathematical model should be made by means of the NN and other AI technology.
On the basis of combining the characteristic of the technics theory and practical devanadium process, this paper conclude the internal affect factor of devanadium process and the main factor of static model. Then the author has constructed the submodel of the time of blowing oxygen and refrigerant, making use of the RBF neural networks. Meantime, applying the genetic algorithm, completed the optimizationof parameter of the number of neuron and spread, and get the model with a better generalization ability. The imitate results show that the algorithm has an ability of fast
convergence speed and strong study power.
At last, this paper develop a software for devanadium static control model by means of object-oriented programming language Visual C++ 6.0.
引文
[1]黄道鑫主编,提钒炼钢，冶金工业出版社，2000:1
[2]汪子和，国外钒钛工业与钡钛磁铁矿的综合利用，重庆科技情报所，1993
[3]古隆建，我国火法提钒技术的进展及现状，钒钛，1992.6,29-39
[4]攀枝花钢铁研究院情报室，国内提钒工艺的现状与发展，提钒文献汇编，1987，1
[5]陈家祥主编，钢铁冶金学，冶金工业出版社：1990:5
[6] 野板康雄编著，舒诒湘译，钢铁工业中的计算机控制，上海科学技术出版社:1982
[7] S.M. Galloway, M.J.Green, S. R. Balajee, Imporvement in Furnace Performance at Inland Steel Company's No 2 BOF Shop Through Models Utilization and Standardization of Operating Practices. STEELMAKING CONFERENCE PROCEEDINGS. 1991,389~396
[8] T Hara,H Nakamura,M Tawara, et al. DEVELOPMENT OF DIRECT TAPPING AT NO 2 STEELMAKING SHOP IN KURE WORKS, STEELMAKING CONFERENCE PROCEEDINGS, 1989, 155-158
[9] A Abbulangelo,M Dalla Rena,M Palcbelli,et.al. BLOWING PATTERN COMPUTERISED CONTROL AT TARANTO STEELSHOP. STEELMAKING CONFERENCE PROCEEDINGS. 19909,337~342
[10] 夏升华，余达太，曹同乐，25t转炉自适应静态模型动态化控制，北京科技大学学报，1998，20，(6)，579~584
[11] KerrTwamura,Masao Furusawa,Masakazu Miyamoto,et al. New Endpoint Control System with Auto-Parameter Turning in BOF. Steelmaking conference proceedings,1995,715-719
[12] D Partington Expert Systems Developments within British Steel Used and Application of Expert Systems and Articial Intelligence,Brussels,Belgium,IISI,1989,15~29
[13]竹中正树，钿中陪男等，NKK技报，1991,(134):23~28
[14]樊俊飞，宝钢转炉吹炼控制模拟在线专家系统，北京科技大学学报1995.17(增刊)，6-11
[15]徐广新，郑德玲，刘宏才，转炉炼钢后期补吹专家系统的建立，冶金自动化，1991，15，(4)：31~34
[16]钢铁研究总院，北京科技大学，转炉炼钢过程控制中的人工智能技术，炼钢，1998(3)，59~63
[17]谢书明，柴天佑,陶钧，一种转炉炼钢动态终点预报的新方法，自动化学报，2001,27,(1),136~139
[18]潘树范，国内外氧气顶吹转炉提钒现状及对攀钢转炉提钒有关问题的探讨，钢铁钒钛，1995.16(1)，6-16
[19]杨素波，戈文荪，王建等，攀钢转炉提钒工艺流程中钒的走向分析，钢铁钒钛，1998.19,(2),20-23
[20]过程辩识，方崇智，萧德云编著，北京,清华大学出版社，1998
[21]雷新泉，雷曼，钢铁研究，转炉炼钢静态模型的初步探讨，钢铁研究，1994，2，8~12
[22]高等过程控制，王桂增，王诗宓编著,清华大学出版社，2002
[23] 渝舒仁，转炉炼钢过程静态控制及其数学模型，炼钢，1995,(6),55~60
[24]Honik K. Approximation Capabilities of Multilayer Feedforwad NetWorks. Neural Networks. 1991,4:551~557
[25]潘正君 ,康立山，陈毓屏编著，演化计算，清华大学出版社，1998
[26]赵新昱，陈文伟，基于遗传建模的公式发现研究，计算机工程与科学，2000，22，(5)，84~87
[27]Vapnik V. The Nature of Statistical Learning Theory,New York:Springer-Verlag,1995
[28]Vapnik V. The Nature of Statistical Learning Theory,New York:Springer-Verlag,1995
[29]Cortes C,Vapnik V.,Support Vector Networks,Machine Learning ,1995:20:273-297
[30]Vapnik V,Golowich S,Smola A, Support vector method for function approximation,regression estimation ,and signal processing ,In:Mozer M,Jordan M,Petsche T(eds). Neural Information Processing Systems,MIT Press,1997,9
[31] Muller K-R,Smola A J,Ratsch G et al. Predictiong time series with support vector machine . In : Proc. Of ICANN'97, Springer Lecture Notes in Computer Science,1997,999~1005
[32]Montazemi A R,Gunpta K M. An adaptive agent for case description in diagnostic CBR system[J]. Journal of Computers in Industry,1996,29(3),209-224
[33]Improving human decision making through case-based decision aiding[J], AI Magazine,1991,12(2):52-68
[34]王永骥，涂健编著，神经网络控制，北京，机械工业出版社，1999
[35]韩力群，人工神经网络在控制中的应用及发展前景，北京轻工业学院学报，1996.14(1),32-36
[36]王耀南著，智能控制系统，长沙，湖南大学出版社，1996
[37]高大启，有教师的线性基本函数前向三层神经网络结构研究，计算机学报，1998.Vol.21,No.1:80-83
[38] 吴斌 吴坚 涂序彦，快速遗传算法研究，电子科技大学学报，1999，Vol.28 No.1：49-53
[39] Robert J S，Rovert J M，Dynamic fuzzy control of genetic algorithm parameter coding，IEEE Trans Syst Man，Cybern，1999,29(3):426-433
[40] 董聪，关仪，遗传算法，大自然探索，1998，16（1）：37-41
[41] 董聪 郭晓华 袁曾任，基于广义遗传算法的全局优化方法，计算机科学，Vol.26 No.6 ,1999：7~10
[42] 李聪 张勇 高智，一种新的聚类算法，模式识别与人工智能，1999（6）：205-208
[43]Baker J E. Adaptive Selection Methods for Genetic Algorithms. ,Grefenstette J J(Ed.).Proc. of the 2nd Int'l. Conf. on Genetic Algorithms. Lawrence Earlbaum Associates,Hilladale,NJ,1985
[2]汪子和，国外钒钛工业与钡钛磁铁矿的综合利用，重庆科技情报所，1993
[3]古隆建，我国火法提钒技术的进展及现状，钒钛，1992.6,29-39
[4]攀枝花钢铁研究院情报室，国内提钒工艺的现状与发展，提钒文献汇编，1987，1
[5]陈家祥主编，钢铁冶金学，冶金工业出版社：1990:5
[6] 野板康雄编著，舒诒湘译，钢铁工业中的计算机控制，上海科学技术出版社:1982
[7] S.M. Galloway, M.J.Green, S. R. Balajee, Imporvement in Furnace Performance at Inland Steel Company's No 2 BOF Shop Through Models Utilization and Standardization of Operating Practices. STEELMAKING CONFERENCE PROCEEDINGS. 1991,389~396
[8] T Hara,H Nakamura,M Tawara, et al. DEVELOPMENT OF DIRECT TAPPING AT NO 2 STEELMAKING SHOP IN KURE WORKS, STEELMAKING CONFERENCE PROCEEDINGS, 1989, 155-158
[9] A Abbulangelo,M Dalla Rena,M Palcbelli,et.al. BLOWING PATTERN COMPUTERISED CONTROL AT TARANTO STEELSHOP. STEELMAKING CONFERENCE PROCEEDINGS. 19909,337~342
[10] 夏升华，余达太，曹同乐，25t转炉自适应静态模型动态化控制，北京科技大学学报，1998，20，(6)，579~584
[11] KerrTwamura,Masao Furusawa,Masakazu Miyamoto,et al. New Endpoint Control System with Auto-Parameter Turning in BOF. Steelmaking conference proceedings,1995,715-719
[12] D Partington Expert Systems Developments within British Steel Used and Application of Expert Systems and Articial Intelligence,Brussels,Belgium,IISI,1989,15~29
[13]竹中正树，钿中陪男等，NKK技报，1991,(134):23~28
[14]樊俊飞，宝钢转炉吹炼控制模拟在线专家系统，北京科技大学学报1995.17(增刊)，6-11
[15]徐广新，郑德玲，刘宏才，转炉炼钢后期补吹专家系统的建立，冶金自动化，1991，15，(4)：31~34
[16]钢铁研究总院，北京科技大学，转炉炼钢过程控制中的人工智能技术，炼钢，1998(3)，59~63
[17]谢书明，柴天佑,陶钧，一种转炉炼钢动态终点预报的新方法，自动化学报，2001,27,(1),136~139
[18]潘树范，国内外氧气顶吹转炉提钒现状及对攀钢转炉提钒有关问题的探讨，钢铁钒钛，1995.16(1)，6-16
[19]杨素波，戈文荪，王建等，攀钢转炉提钒工艺流程中钒的走向分析，钢铁钒钛，1998.19,(2),20-23
[20]过程辩识，方崇智，萧德云编著，北京,清华大学出版社，1998
[21]雷新泉，雷曼，钢铁研究，转炉炼钢静态模型的初步探讨，钢铁研究，1994，2，8~12
[22]高等过程控制，王桂增，王诗宓编著,清华大学出版社，2002
[23] 渝舒仁，转炉炼钢过程静态控制及其数学模型，炼钢，1995,(6),55~60
[24]Honik K. Approximation Capabilities of Multilayer Feedforwad NetWorks. Neural Networks. 1991,4:551~557
[25]潘正君 ,康立山，陈毓屏编著，演化计算，清华大学出版社，1998
[26]赵新昱，陈文伟，基于遗传建模的公式发现研究，计算机工程与科学，2000，22，(5)，84~87
[27]Vapnik V. The Nature of Statistical Learning Theory,New York:Springer-Verlag,1995
[28]Vapnik V. The Nature of Statistical Learning Theory,New York:Springer-Verlag,1995
[29]Cortes C,Vapnik V.,Support Vector Networks,Machine Learning ,1995:20:273-297
[30]Vapnik V,Golowich S,Smola A, Support vector method for function approximation,regression estimation ,and signal processing ,In:Mozer M,Jordan M,Petsche T(eds). Neural Information Processing Systems,MIT Press,1997,9
[31] Muller K-R,Smola A J,Ratsch G et al. Predictiong time series with support vector machine . In : Proc. Of ICANN'97, Springer Lecture Notes in Computer Science,1997,999~1005
[32]Montazemi A R,Gunpta K M. An adaptive agent for case description in diagnostic CBR system[J]. Journal of Computers in Industry,1996,29(3),209-224
[33]Improving human decision making through case-based decision aiding[J], AI Magazine,1991,12(2):52-68
[34]王永骥，涂健编著，神经网络控制，北京，机械工业出版社，1999
[35]韩力群，人工神经网络在控制中的应用及发展前景，北京轻工业学院学报，1996.14(1),32-36
[36]王耀南著，智能控制系统，长沙，湖南大学出版社，1996
[37]高大启，有教师的线性基本函数前向三层神经网络结构研究，计算机学报，1998.Vol.21,No.1:80-83
[38] 吴斌 吴坚 涂序彦，快速遗传算法研究，电子科技大学学报，1999，Vol.28 No.1：49-53
[39] Robert J S，Rovert J M，Dynamic fuzzy control of genetic algorithm parameter coding，IEEE Trans Syst Man，Cybern，1999,29(3):426-433
[40] 董聪，关仪，遗传算法，大自然探索，1998，16（1）：37-41
[41] 董聪 郭晓华 袁曾任，基于广义遗传算法的全局优化方法，计算机科学，Vol.26 No.6 ,1999：7~10
[42] 李聪 张勇 高智，一种新的聚类算法，模式识别与人工智能，1999（6）：205-208
[43]Baker J E. Adaptive Selection Methods for Genetic Algorithms. ,Grefenstette J J(Ed.).Proc. of the 2nd Int'l. Conf. on Genetic Algorithms. Lawrence Earlbaum Associates,Hilladale,NJ,1985