钢铁焦炉煤气产消及柜位预测方法与应用
|
摘要
钢铁焦炉煤气系统是企业副产煤气二次能源的主要组成部分。对焦炉煤气发生量和消耗量及煤气柜位的准确预测,可以改变目前调度人员仅凭经验实现焦炉煤气系统平衡的状态,为制定合理煤气调度方案提供科学指导,从而减少煤气放散损失,降低钢铁企业的能耗。
本文通过分析焦炉煤气的发生机理和消耗特性,确定了影响煤气柜柜位变化的主要煤气用户,将焦炉煤气的产消预测归结为一类基于时间序列的预测问题,将煤气柜柜位预测归结为回归预测问题。建立了相应的基于最小二乘支持向量机的产消预测模型和柜位预测模型,并设计了在线学习算法和贝叶斯优化法循环构建和优化预测模型,加快了建模时间,同时提高了预测精度。现场实际数据预测结果表明所建模型在小样本和随机噪声数据环境下能保持很高的预测精度,与其它预测模型相比,适合于钢铁企业的焦炉煤气发生量实时在线预测。
针对宝钢能源调度中心调度人员的实际需要,基于本文建立的产消预测模型和柜位预测模型,开发了宝钢焦炉煤气产消预测和调度系统,目前该系统已经投入到宝钢能源中心进行试运行,为调度人员的能源配置及调度过程提供定量的支持。
Cove oven gas (COG) as a class of byproduct gas is a significant part of the secondary fuel in steel industry. Precisely forecasting the holder level and generation-consumption demand of the COG system can provide the scheduling workers with the scientific and reasonable guidance for balancing the whole gas system, reducing the waste of gas diffusion and decreasing the energy consumption of the steel enterprise.
Based on an analysis of generation mechanism and the consumption characteristics of COG system, the main system units (COG users) that affect the COG gas holder level are determined. The generation and consumption of the COG are considered as a class of prediction problem based on the time series. The prediction of gas holder level is considered as a regression problem. A generation and consumption model and a gas holder level model are established based on the least square support vector machine (LSSVM) for COG system in this paper. An on-line learning algorithm and the circular Bayesian optimization are modeled in order to accelerate the model establishment and enhance the prediction precision. The practical production data from Shanghai BaoSteel is used to verify the proposed model and algorithm. The results demonstrate that the prediction precision can be greatly guaranteed under the circumstance of small training sample and high random noises. In addition, the comparison to some other prediction model also shows that the presented approach accommodates to the on-line prediction process of COG system in steel industry.
Combined with the requirement of the Energy Center in Baosteel, the generation-consumption prediction and scheduling system for COG system is developed on the basis of the proposed generation-consumption model and holder level prediction model. At present, this system has been run for test in the Energy Center of Baosteel and provides with an effective guidance for the energy scheduling workers.
本文通过分析焦炉煤气的发生机理和消耗特性,确定了影响煤气柜柜位变化的主要煤气用户,将焦炉煤气的产消预测归结为一类基于时间序列的预测问题,将煤气柜柜位预测归结为回归预测问题。建立了相应的基于最小二乘支持向量机的产消预测模型和柜位预测模型,并设计了在线学习算法和贝叶斯优化法循环构建和优化预测模型,加快了建模时间,同时提高了预测精度。现场实际数据预测结果表明所建模型在小样本和随机噪声数据环境下能保持很高的预测精度,与其它预测模型相比,适合于钢铁企业的焦炉煤气发生量实时在线预测。
针对宝钢能源调度中心调度人员的实际需要,基于本文建立的产消预测模型和柜位预测模型,开发了宝钢焦炉煤气产消预测和调度系统,目前该系统已经投入到宝钢能源中心进行试运行,为调度人员的能源配置及调度过程提供定量的支持。
Cove oven gas (COG) as a class of byproduct gas is a significant part of the secondary fuel in steel industry. Precisely forecasting the holder level and generation-consumption demand of the COG system can provide the scheduling workers with the scientific and reasonable guidance for balancing the whole gas system, reducing the waste of gas diffusion and decreasing the energy consumption of the steel enterprise.
Based on an analysis of generation mechanism and the consumption characteristics of COG system, the main system units (COG users) that affect the COG gas holder level are determined. The generation and consumption of the COG are considered as a class of prediction problem based on the time series. The prediction of gas holder level is considered as a regression problem. A generation and consumption model and a gas holder level model are established based on the least square support vector machine (LSSVM) for COG system in this paper. An on-line learning algorithm and the circular Bayesian optimization are modeled in order to accelerate the model establishment and enhance the prediction precision. The practical production data from Shanghai BaoSteel is used to verify the proposed model and algorithm. The results demonstrate that the prediction precision can be greatly guaranteed under the circumstance of small training sample and high random noises. In addition, the comparison to some other prediction model also shows that the presented approach accommodates to the on-line prediction process of COG system in steel industry.
Combined with the requirement of the Energy Center in Baosteel, the generation-consumption prediction and scheduling system for COG system is developed on the basis of the proposed generation-consumption model and holder level prediction model. At present, this system has been run for test in the Energy Center of Baosteel and provides with an effective guidance for the energy scheduling workers.
引文
[1]李新创.中国钢铁工业结构调整及对原材料需求展望[J].钢铁,2006,6(12):2-17.
[2]刘渺.钢铁企业主工序分厂煤气量[D].长沙:中南大学,2006.
[3]殷瑞钰,蔡九菊.钢厂生产流程与大气排放[J].钢铁,1999,34(5):61-65.
[4]娄湖山.国内外钢铁工业能耗现状和发展趋势及节能对策[J].冶金能源,2007,26(2):7-11.
[5]田敬龙.中国十大钢铁企业能耗分析及节能工作建议[J].冶金能源,2007,26(6):3-7.
[6]郑文华,张兴柱.焦炉煤气的使用现状与应用前景[J].燃料与化工,2004,35(4):1-3.
[7]王太炎.焦炉煤气开发利用的问题与途径[J].燃料与化工,2004,35(6):1-3.
[8]杨晓,张玲.钢铁工业能源消耗和二次能源利用途径及对策[J].钢铁,2000,35(20):64-67.
[9]张琦,蔡九菊,杜涛.钢铁联合企业煤气系统优化利用[J].冶金能源,2005,24(5):9-16.
[10]Mousa S,Mohsen and Bilal A,Akash.Energy analysis of the steel making industry[J].International Journal of Energy Research,1998,22(12):1049-1054.
[11]Naoki H,~oshitsugu W,Takeyoshi K,et al.Minimizing energy consumption in Industries by cascade use of waste energy[J].IEEE Transactions on Energy Conversion,1999,3(14):795-801.
[12]S.R.Valsalam,V.Muralidharan and N.Krishnan.Implementation of energy management system for an integrated steel plant[J].Proceedings of Energy Management and Power Delivery,1998,2:661-666.
[13]王鼎,方结.能源中心在宝钢能源生产中的作用和发展趋势[J].中国冶金,2005,1(86):22-23.
[14]孙贻公.对大型钢铁联合企业煤气平衡问题的探讨[J].钢铁,2003,38(6):59-64.
[15]赵立合,林秀贞,朱和杰,等.钢铁企业煤气系统优化管理模型及其应用[J].冶金能源,1994,13(6):10一13.
[16]Ming D T,Li J,Yin Y X.Research of gas optimal scheduling model in iron and steel enterprises[J].Computer Engineering and Design,2008,29(6):1575-1578.
[17]唐万梅.几个预测方法及模型研究[D].包头:内蒙古大学,2006.
[18]John Fox.Applied Regression Analysis,Linear Models,and Related Methods[M].SAGE,1997.
[19]贺静,韦钢,熊玲玲.负荷预测线性回归分析法的模糊改进阴[J].华东电力,2003年第11期:21-23.
[20]George Edward,Pelham Box,Gwilym Jenkins.Time series analysis,forecasting and control[M].Holden-Day,Incorporated,1990.
[21]杨金芳,翟永杰,王东风,等.基于支持向量回归的时间序列预测[J].中国电机工程学报,2005,25(7)110-114.
[22]李伟,韩力.组合灰色预测模型在电力负荷预测中的应用明[J].重庆大学学报,2004,27(1):36-39.
[23]刘思峰,郭天榜,党耀国.灰色系统理论及其应用[M].北京科学出版社:北京,1999.
[24]邓聚龙.灰色系统理论教程[M].武汉:华中科技大学,1990.
[25]Christopher M.Bishop.Neural networks for pattern recognition[M].Oxford University Press,1995.
[26]刘颖,赵瑁,王伟,等.基于数据的改进回声状态网络在高炉煤气发生量预测中的应用[J].自动化学报,2009,35(6):73 1-738.
[27]Suykens J A K.Least squares support vector machines for classification and nonlinear modeling[J].Neural network world,2000,10:29-48.
[28]S.R.Gunn.Support Vector Machines for Classification and Regression[R].Technical Report,Image Speech and Intelligent Systems Research Group,University of Southampton,1997.
[29]姚昭章.炼焦学[M].北京:冶金工业出版社,1983,84-98.
[30]张应良.能源介质的综合调整法[J].能源技术,2000,6(2):115-117.
[31]张建良,王好.钢铁企业煤气系统的优化利用模型[J].包头钢铁学院学报,2002,21(3):280-282.
[32]张琦,蔡九菊,杜涛,等.钢铁企业煤气平衡问题的探讨[J].节能,2004,12:9-11.
[33]Xiao X Y,Ge J,He D SH.Combination Method of Mid-Long Term Load Forecasting Based on Support Vector Machine[J].Proceedings of the CSU-EPSA,2008,20(1)84-87.
[34]张国华.基于神经网络的短期负荷预测[D].北京:华北电力大学,2000.
[35]赵登福,王蒙,张讲社,等.基支撑向量机方法的短期负荷预测[J].中国电机工程学报,2002,22(4):26-30.
[36]Suykens J A K,De Brabanter J,Lukas L,et al.Weighted least squares support vector machines:robustness and sparse approximation[J].Neurocomputing,2002,48(14):85-105.
[37]Flandrin P,Rilling G,Goncalves P.Empirical mode decomposition as a filter bank[J].IEEE Signal Processing Letters,2004,11(2):112-114.
[38]Tony Van Gestel,Suykens J A K,et al.Financial time series prediction using least squares support vector machines within the evidence framework[J].IEEE transactions on neural networks,2001,12(4):809-821.
[39]Marcelo Espinoza,Suykens J A K,Bart De Moor.Fixed-size least squares support vector machines:a large scale application in electrical load forecasting[J].Computational Management Science,2006,3(2):113-129.
[40]Li L J,Su H Y,Chu J.Generalized predictive control with online least squares support vector machines[3].ACTA AUTOMATICA SINICA,2007,33(11):1182-1188.
[41]Cawley G C,Talbot N L C.Fast exact leave-one-out cross-validation of sparse least squares support vector machines[J].Neural Networks,2004,10(17):1467-1475.
[42]Tao S H,Chen D Z,Hu W Y.Gradient algorithm for selecting hyper parameters of LSSVM in process modeling[J].Journal of Chemical Industry and Engineering(China),2007,6(58) 1515-1517.
[43]Efron,B,Morris C N.Empirical Bayes on Vector Observations:h Extension of Stein's Method[J].Biometrika,1972,59:335-347.
[44]陈磊,张士乔.基于贝叶斯最小二乘支持向量机的时用水量预测模型[J].天津大学学报,2006,39(9):1037-1042.
[45]Guo B,Liu H P,Wang L.Method for Selecting Parameters of Least Squares Support Vector Machines and Application[J].Journal of System Simulation,2006,18(7):2033-2036.
[46]Chi M V,Wong P K,Li Y P.Prediction of automotive engine powerand torque using least squares support vector machines and Bayesian inference[J].Engineering Applications of Artificial Intelligence,2006,19:277-287.
[2]刘渺.钢铁企业主工序分厂煤气量[D].长沙:中南大学,2006.
[3]殷瑞钰,蔡九菊.钢厂生产流程与大气排放[J].钢铁,1999,34(5):61-65.
[4]娄湖山.国内外钢铁工业能耗现状和发展趋势及节能对策[J].冶金能源,2007,26(2):7-11.
[5]田敬龙.中国十大钢铁企业能耗分析及节能工作建议[J].冶金能源,2007,26(6):3-7.
[6]郑文华,张兴柱.焦炉煤气的使用现状与应用前景[J].燃料与化工,2004,35(4):1-3.
[7]王太炎.焦炉煤气开发利用的问题与途径[J].燃料与化工,2004,35(6):1-3.
[8]杨晓,张玲.钢铁工业能源消耗和二次能源利用途径及对策[J].钢铁,2000,35(20):64-67.
[9]张琦,蔡九菊,杜涛.钢铁联合企业煤气系统优化利用[J].冶金能源,2005,24(5):9-16.
[10]Mousa S,Mohsen and Bilal A,Akash.Energy analysis of the steel making industry[J].International Journal of Energy Research,1998,22(12):1049-1054.
[11]Naoki H,~oshitsugu W,Takeyoshi K,et al.Minimizing energy consumption in Industries by cascade use of waste energy[J].IEEE Transactions on Energy Conversion,1999,3(14):795-801.
[12]S.R.Valsalam,V.Muralidharan and N.Krishnan.Implementation of energy management system for an integrated steel plant[J].Proceedings of Energy Management and Power Delivery,1998,2:661-666.
[13]王鼎,方结.能源中心在宝钢能源生产中的作用和发展趋势[J].中国冶金,2005,1(86):22-23.
[14]孙贻公.对大型钢铁联合企业煤气平衡问题的探讨[J].钢铁,2003,38(6):59-64.
[15]赵立合,林秀贞,朱和杰,等.钢铁企业煤气系统优化管理模型及其应用[J].冶金能源,1994,13(6):10一13.
[16]Ming D T,Li J,Yin Y X.Research of gas optimal scheduling model in iron and steel enterprises[J].Computer Engineering and Design,2008,29(6):1575-1578.
[17]唐万梅.几个预测方法及模型研究[D].包头:内蒙古大学,2006.
[18]John Fox.Applied Regression Analysis,Linear Models,and Related Methods[M].SAGE,1997.
[19]贺静,韦钢,熊玲玲.负荷预测线性回归分析法的模糊改进阴[J].华东电力,2003年第11期:21-23.
[20]George Edward,Pelham Box,Gwilym Jenkins.Time series analysis,forecasting and control[M].Holden-Day,Incorporated,1990.
[21]杨金芳,翟永杰,王东风,等.基于支持向量回归的时间序列预测[J].中国电机工程学报,2005,25(7)110-114.
[22]李伟,韩力.组合灰色预测模型在电力负荷预测中的应用明[J].重庆大学学报,2004,27(1):36-39.
[23]刘思峰,郭天榜,党耀国.灰色系统理论及其应用[M].北京科学出版社:北京,1999.
[24]邓聚龙.灰色系统理论教程[M].武汉:华中科技大学,1990.
[25]Christopher M.Bishop.Neural networks for pattern recognition[M].Oxford University Press,1995.
[26]刘颖,赵瑁,王伟,等.基于数据的改进回声状态网络在高炉煤气发生量预测中的应用[J].自动化学报,2009,35(6):73 1-738.
[27]Suykens J A K.Least squares support vector machines for classification and nonlinear modeling[J].Neural network world,2000,10:29-48.
[28]S.R.Gunn.Support Vector Machines for Classification and Regression[R].Technical Report,Image Speech and Intelligent Systems Research Group,University of Southampton,1997.
[29]姚昭章.炼焦学[M].北京:冶金工业出版社,1983,84-98.
[30]张应良.能源介质的综合调整法[J].能源技术,2000,6(2):115-117.
[31]张建良,王好.钢铁企业煤气系统的优化利用模型[J].包头钢铁学院学报,2002,21(3):280-282.
[32]张琦,蔡九菊,杜涛,等.钢铁企业煤气平衡问题的探讨[J].节能,2004,12:9-11.
[33]Xiao X Y,Ge J,He D SH.Combination Method of Mid-Long Term Load Forecasting Based on Support Vector Machine[J].Proceedings of the CSU-EPSA,2008,20(1)84-87.
[34]张国华.基于神经网络的短期负荷预测[D].北京:华北电力大学,2000.
[35]赵登福,王蒙,张讲社,等.基支撑向量机方法的短期负荷预测[J].中国电机工程学报,2002,22(4):26-30.
[36]Suykens J A K,De Brabanter J,Lukas L,et al.Weighted least squares support vector machines:robustness and sparse approximation[J].Neurocomputing,2002,48(14):85-105.
[37]Flandrin P,Rilling G,Goncalves P.Empirical mode decomposition as a filter bank[J].IEEE Signal Processing Letters,2004,11(2):112-114.
[38]Tony Van Gestel,Suykens J A K,et al.Financial time series prediction using least squares support vector machines within the evidence framework[J].IEEE transactions on neural networks,2001,12(4):809-821.
[39]Marcelo Espinoza,Suykens J A K,Bart De Moor.Fixed-size least squares support vector machines:a large scale application in electrical load forecasting[J].Computational Management Science,2006,3(2):113-129.
[40]Li L J,Su H Y,Chu J.Generalized predictive control with online least squares support vector machines[3].ACTA AUTOMATICA SINICA,2007,33(11):1182-1188.
[41]Cawley G C,Talbot N L C.Fast exact leave-one-out cross-validation of sparse least squares support vector machines[J].Neural Networks,2004,10(17):1467-1475.
[42]Tao S H,Chen D Z,Hu W Y.Gradient algorithm for selecting hyper parameters of LSSVM in process modeling[J].Journal of Chemical Industry and Engineering(China),2007,6(58) 1515-1517.
[43]Efron,B,Morris C N.Empirical Bayes on Vector Observations:h Extension of Stein's Method[J].Biometrika,1972,59:335-347.
[44]陈磊,张士乔.基于贝叶斯最小二乘支持向量机的时用水量预测模型[J].天津大学学报,2006,39(9):1037-1042.
[45]Guo B,Liu H P,Wang L.Method for Selecting Parameters of Least Squares Support Vector Machines and Application[J].Journal of System Simulation,2006,18(7):2033-2036.
[46]Chi M V,Wong P K,Li Y P.Prediction of automotive engine powerand torque using least squares support vector machines and Bayesian inference[J].Engineering Applications of Artificial Intelligence,2006,19:277-287.