基于人工神经网络的铁水预处理终点硫含量预报模型
详细信息    本馆镜像全文|  推荐本文 |  |   获取CNKI官网全文
摘要
随着冶金工业的发展和钢质量的不断提高,铁水预脱硫成为钢铁生产工艺流程中的一项重要任务。为了实现铁水预处理工艺过程快节奏、高效率化的生产发展需求,前人提出了利用计算机模型进行铁水预处理终点硫含量预报的方法。但预报终点硫含量的过程是一个非常复杂的工艺过程,应用传统的工艺理论建模已难以适应其多参数、非线性和高度不确定对象的特点,因此近年来多采用人工智能的方法来进行预报。
     课题以梅山钢铁公司(以下简称梅钢)和本溪钢铁公司(以下简称本钢)的铁水预处理生产工艺为研究背景,采用改进的BP算法,应用Visual Basic 6.0高级程序语言进行程序设计,建立铁水预处理终点硫含量预报模型。模型建立过程中,针对BP网络迭代次数多、收敛速度慢等问题对标准BP算法进行了分析和改进,得到了适于本模型的改进型BP算法。对模型中各个参数的选择做了较详细的选择分析,从热力学和动力学的角度出发,结合现场数据情况,深入考察了影响铁水预处理终点硫含量的各种因素,确定了模型的网络结构及输入、输出参数。
     用梅钢的1154炉数据和本钢的1900炉数据作为模型的训练样本,另外,再分别随机选取100炉数据作为模型测试样本,分别对模型进行了训练和测试。然后,对产生误差的原因以及模型各个输入参数与终点硫含量的关系进行了分析和讨论。
     课题得到的主要结论如下:
     (1)提出了采用自适应调整学习率、增加动量项和最大误差学习法的适合本课题使用的改进BP算法。其中,新提出的自适应调整学习率改进方法如下:
     (2)确定模型输入参数为:铁水温度、铁水重量、镁粉耗量、石灰粉耗量、初始硫含量;模型的输出参数为:终点硫含量;
     (3)改进的BP算法比标准BP算法预报误差≤0.003%的精度提高28%;
     (4)梅钢模型的网络结构为5-14-1结构,动量项为0.6。本钢模型的网络结构为5-10-1结构,动量项为0.7。输入输出数据归一化范围均为[0.2,0.8]区间;
     (5)梅钢模型的预报结果有19%的炉次预报值与实际值完全一致,有90%的炉次误差≤0.003%,达到96%的炉次误差≤0.005%,平均误差为0.0017%;
With the development of metallurgical industry and the improvement of steel quality, pre-desulfurization of hot metal has become an important task for steel production. In order to meet the demand of hot metal pretreatment processing with the character of fast rhythm and high efficiency, the predicting of the final sulfur content with computer model are put forward by the predecessors. However, it is a complex process to predict the final sulfur content. It is unsuitable for the features of multiparameter, nonlinear and uncertainty to modeling with traditional theroy model, and then the artifical intelligence method is applied to predict recently.
    Based on the productive practice of Meishan Steel Co. Ltd. and Benxi Steel Co. Ltd., adopted the improved BP algorithm, and used Visual Basic 6.0 programme software, the prediction model of final sulfur content during hot metal pretreatment processing is established. During modeling process, normal BP algorithm is analysed and improved for overcome its disadvantages of overmuch iterative repetition and slow convergence. All kinds of parameters in the model are elaborated. In the view of thermodynamic, kinetics and combining the characteristic of the field datas, the factors affecting final sulfur content during hot metal pretreatment processing are detailedly investigated. At the same time, the network configuration, input and output parameters are established.
    Training samples of Meishan Iron and Steel Co. Ltd. are 1154 heats, and which are 1900 heats for Benxi Iron and Steel Co. Ltd. 100 heats datas are randomly selected as the test samples respectively, and which are different from the above heats. The model is seperately trained and tested using the selected samples. And then the reasons resulting in error are analysed and discussed. The following conclusions are drawn:
    (1) The improved BP algorithm which is adaptive to this subject is put forward by adjusting study rate, adding momentum coefficient and employing the learning method of maximal error. The new study rate is as follows:
引文
1.欧阳守忠.铁水脱硫预处理方法及其有关新技术的发展[J],炼钢,1995,11(3):61.
    2.杨天钧,高征铠.铁水炉外脱硫的新进展[J],钢铁,1999,34(1):65-69.
    3.吴义生,高广才,宫玉秀.国内外铁水脱硫预处理技术发展概况[J],山东冶金,2000,22(2):8-11.
    4. Grosjean J C, Reboul J P, Bauler C, et al. Industrial studies on hot metal pretreatment in France[J], Steel Times, 1987, 215(11): 542, 544-545.
    5. Selin, Roger. Uses of lime-based fluxes for simultaneous removal of phosphorus and sulphur in hot metal pretreatment[J], Scandinavian Journal of Metallurgy, 1990, 19(3): 98-109.
    6. Vargas-Ramirez, M. Hot metal pretreatment by powder injection of lime-based reagents[J], Steel Research, 2001, 72(5): 173-181.
    7.杨世山.铁水预处理工艺、设备及操作[J],炼钢,2000,16(5):14-15.
    8.韦斯伯格S,王静龙译.应用线性回归[M],北京:中国统计出版社,1998,1.
    9.余永军.喷吹CaO-Mg粉剂铁水脱硫工艺分析及其参数优化[D],沈阳:东北大学,2000.
    10.王贵平,张华书,王建昌.太钢铁水炉外喷粉脱硫经验模型[A],全国铁水预处理技术研讨会文集[C],大连:中国金属学会,2003,99-102.
    11.印鉴,刘星成,汤庸.专家系统原理与编程[M],北京:机械工业出版社,2000,2.
    12.李丽,孙云兰,陈雪波.一个确定铁水脱硫剂喷吹量的专家系统[J],鞍山科技大学学报,2004,27(2):120-123.
    13.周继成.人工神经网络—第六代计算机的实现[M],北京:科学普及出版社,1993,25-26.
    14.李治友,陈才,曹长修.一种基于改进的RBF神经网络的铁水脱硫预报模型[J],重庆大学学报,2003,26(9):120-122.
    15. Amlan D, Mavoori H, Prem K K, et al. Adaptive neural net models for desulphurization of hot metal and steel[J], Steel Research, 1994, 65 (11): 466-471.
    16.王华秋,曹长修,张邦礼.增量式遗传RBF神经网络在铁水脱硫预处理中的应用[J],信息与控制,2004,33(1):89-92.
    17.张苹,韩大平,郝晓静,等.BP算法的模糊神经网络及烧结终点辅助预测模型[J],材料与冶金学报,2004,3(2):157-160.
    18.程武山.基于遗传神经网络的烧结终点预测系统[J],烧结球团,2004,19(5):18-22.
    19.李桃,冯其明,范晓慧,等.基于自组织神经网络的烧结终点自适应预报系统的开发[J],计算机工程与应用,2001,(6):127-129.
    20.李俊国,闫小林.高炉铁水含硅量神经网络预测模型[J],河北理工学院学报,2003,24(3):17-22.21.杨尚宝,杨天钧,董一诚.铁水含硅量预报神经网络模型[J],北京科技大学学报,1995,17(6):524-528.
    22.李家新,周莉英,唐成润.神经网络在梅山高炉铁水硅含量预报中的应用[J],钢铁,2001,36(5):14-17.
    23.谢书明,陶钧,柴天佑.基于神经网络的转炉炼钢终点控制[J],控制理论与应用,2003,20(6):903-907.
    24.杨立红,刘浏,何平.基于自适应模糊神经网络系统的转炉终点磷的预报控制模型[J],钢铁研究学报,2002,14(4):47-51.
    25.杨立红,刘浏,何平.转炉冶炼终点锰成分的预报模型[J],炼钢,2003,19(1):10-13.
    26.房荣波,魏元,杨海峰.转炉钢水出钢温度的预测[J],鞍钢技术,2002,(4):22-26.
    27.屠海,洪新,郑少波,等.转炉炼钢终点锰磷动态控制技术的开发[J],上海金属,2002,24(2):27-30.
    28.刘锟,刘浏,何平,等.增量神经网络模型预报100t电弧炉终点碳、磷和温度的应用[J],特殊钢,2004,25(3):40-41.
    29.张兴.神经网络在50tDC电弧炉炼钢中的运用[J],特殊钢,2002,23(3):38-41.
    30.李亮,姜周华,王文忠,等.应用神经网络技术预报VD炉终点钢水温度[J],钢铁研究学报,2003,15(3):56-59.
    31.杨遴杰,陈伟庆,于平,等.LF-VD-CC钢液温度预报[J],钢铁,2000,35(1):13-16.
    32. Zhang Chunxia, Wang Baojun, Zhou Shiguang, et al. Hybrid neural network model for RH vacuum refining process control[J], Iron and Steel Research International, 2004, 11(1): 12-16.
    33. Wildberger, Martin A, Hickok, et al. Introduction to neural networks[J], Simulation Series, 1989, 20(4): 227-232.
    34. Zurada Jacek M. Introduction to Artificial Neural Systems[M], New York: West Publishing Company, 1992, 145.
    35. Gallent Stephen I. Neural Network Learning and Expert Systems[M], London: The MIT Press, 1993, 56-57.
    36. Simpson Patrick K. Artificial Neural Systems Foundation. Paradigms[M], New York: Pergamon Press, 1990, 25-26.
    37.袁曾任.人工神经元网络及其应用[M],北京:清华大学出版社,1999,1-3.
    38.胡守仁.神经网络应用技术[M],长沙:国防科技大学出版社,1993,22-23.
    39.郑君里,杨行峻.人工神经网络[M],北京:高等教育出版社,1992,18-19.
    40.Martin T H,Howard B D,Mark H B.神经网络设计[M],北京:机械工业出版社,2002,197-258.
    41.蒋中礼.人工神经网络导论[M],北京:高等教育出版社,2001,41-45.42.王科俊,王克成.神经网络建模、预报与控制[M],哈尔滨:哈尔滨工程大学出版社,1996,44-47.
    43. Yu X H, Chen G A, Cheng S X. Dynamic learning rate optimization of the back propagation algorithm[J], IEEE Transactions on Neural Networks, 1995, 6(3): 669-677.
    44. Jin L, Gupta M M. Stable dynamic backpropagation learning in recurrent neural networks[J], IEEE Transactions on Neural Networks, 1999, 10(6): 1321-1334.
    45.焦李成.神经网络系统理论[M],西安:西安电子科技大学出版社,1990,65-73.
    46.赵启林,卓家寿.BP网络的最大误差学习算法[J],河海大学学报,2000,28(1):113-115.
    47. Liu C C, Chen F C. Adaptive control of non-linear continuous-time systems using neural networks-general relative degree and MIMO case[J], Int. J. Control, 1993, 58(2): 317-335.
    48. Chen F C, Liu C. Adaptively controlling nonlinear continuous-time systems using neural networks[A], American Control Conference[C], 1992, 46-50.
    49.梁连科,车荫昌,杨怀,等.冶金热力学及动力学[M],沈阳:东北工学院出版社,1990,202-203.
    50.陈新民,陈启元.冶金热力学导论[M],北京:冶金工业出版社,1986,271-291.
    51. Turkdogan E T. Physical chemistry of high temperature technology[M], New York: Academic Press, 1980, 5.
    52. Hecht-Nielsen, Robert. Theory of the backpropagation neural network[J], Neural Networks, 1988, 1 (1): 445.
    53. Hickok, Kenneth A. Introduction to neural networks[J], Simulation Series, 1989, 20(4): 227-232.
    52.韩力群.人工神经网络理论、设计及应用—人工神经细胞、人工神经网络和人工神经系统[M],北京:化学工业出版社,2002,55-56.
    53.张文鸽,吴泽宁,逯洪波.BP神经网络的改进及其应用[J],河南科学,2003,21(2):202-206.
    54.张荣生.炼钢生产中的脱硫[M],北京:冶金工业出版社,1986,42-45.

© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号

地址:北京市海淀区学院路29号 邮编:100083

电话:办公室:(+86 10)66554848;文献借阅、咨询服务、科技查新:66554700