醋酸乙烯聚合率软测量与预测控制方法研究
|
摘要
醋酸乙烯聚合率是聚乙烯醇生产过程中重要的质量参数,它对生产过程的安全性、经济性和产品品质具有重要影响,但它目前主要靠人工采样离线检测、操作人员依据生产经验进行调控。本文对醋酸乙烯聚合率的在线检测和优化控制进行了探索性研究,主要探讨了两个方面的问题:采用软测量技术实现聚合率的在线估计以及采用预测控制的方法对聚合率进行控制。本文所做的具体研究如下:
1、在了解醋酸乙烯聚合过程工艺机理基础之上,分析了影响醋酸乙烯聚合率的因素,采用了RBF神经网络、最小二乘支持向量机、增量最小二乘支持向量机三种方法建立醋酸乙烯聚合率软测量模型,并利用工业现场的实际测量数据对这几种方法进行了仿真研究和比较。仿真结果表明,增量最小二乘支持向量机方法能根据误差要求调整训练目标,所建模型表现出很高的拟合精度和预测精度,泛化能力强,而且具有在线学习功能,比较适合用于醋酸乙烯聚合率的在线估计。
2、针对聚合反应过程具有非线性、时变、有噪声干扰和和滞后等特点,常规控制方法很难得到令人满意的效果,将预测控制中的隐式广义预测控制算法应用到聚合率控制中。这类算法对模型的精度要求不高,另外,由于在优化中引入了多步预测思想,使其抗扰动及时延变化等能力显著提高。因此本文采用了该算法进行醋酸乙烯聚合率控制系统的设计,并用MATLAB对该算法进行了仿真研究。仿真结果表明该隐式广义预测算法动态响应快,跟踪效果好,能够获得较好的控制效果,并通过仿真分析了隐式广义预测控制中的主要参数对系统性能的影响。
研究结果对醋酸乙烯聚合率在线测量和优化控制具有重要意义。
The vinyl acetate(VAC) polymerization rate is an important quality parameters in the process of the Poly Vinyl Alcohol production, it has significant influence on security, economy of production process and product quality. At present, it mainly relies on manual sampling off-line detection and is controlled by the operators’experiences. This paper mainly studies two problems: on-line estimation of VAC polymerization rate by using soft-sensing technique and optimizing control of VAC polymerization rate by using the method of predictive control. The concrete research works are as follows:
1. On the basis of comprehension of VAC polymerization process mechanics, various factors which affect VAC polymerization rate are analyzed and the soft-sensing model of VAC polymerization rate is established by using three methods including RBF neural network, Least Squares Support Vector Machine(LS-SVM) and Incremental Least Squares Support Vector Machines (ILS-SVM). Then the simulation based on the practical industrial data is researched and compared. The simulation results show that ILS-SVM can adjust training aims according to the error requirement and possesses high fitting and prediction precision, it has good abilities of generalization and the function of studying on line and is fit for on-line estimation of VAC polymerization rate.
2. The process of VAC polymerization has the characteristics such as nonlinearity, time-varying, disturbance and time-delay, the traditional control strategy can’t acquire satisfactory result, so the implicit generalized predictive control(GPC) algorithm is applied in the control of polymerization rate. This algorithm doesn’t depend on exact model, and its ability of disturbance rejection and time-varying restraining has been enhanced greatly with the multi-step forecasting concept. So, the implicit GPC algorithm is adopted to design for the control system of VAC polymerization rate and simulation research is done with MATLAB. The simulation results show that the implicit GPC algorithm can obtain good control performance in both the dynamic response and tracking performance. Finally, the influence of main parameters of the implicit GPC algorithm is analyzed through simulation.
Results shows that it is of importance to online prediction and optimizing control of VAC polymerization rate.
1、在了解醋酸乙烯聚合过程工艺机理基础之上,分析了影响醋酸乙烯聚合率的因素,采用了RBF神经网络、最小二乘支持向量机、增量最小二乘支持向量机三种方法建立醋酸乙烯聚合率软测量模型,并利用工业现场的实际测量数据对这几种方法进行了仿真研究和比较。仿真结果表明,增量最小二乘支持向量机方法能根据误差要求调整训练目标,所建模型表现出很高的拟合精度和预测精度,泛化能力强,而且具有在线学习功能,比较适合用于醋酸乙烯聚合率的在线估计。
2、针对聚合反应过程具有非线性、时变、有噪声干扰和和滞后等特点,常规控制方法很难得到令人满意的效果,将预测控制中的隐式广义预测控制算法应用到聚合率控制中。这类算法对模型的精度要求不高,另外,由于在优化中引入了多步预测思想,使其抗扰动及时延变化等能力显著提高。因此本文采用了该算法进行醋酸乙烯聚合率控制系统的设计,并用MATLAB对该算法进行了仿真研究。仿真结果表明该隐式广义预测算法动态响应快,跟踪效果好,能够获得较好的控制效果,并通过仿真分析了隐式广义预测控制中的主要参数对系统性能的影响。
研究结果对醋酸乙烯聚合率在线测量和优化控制具有重要意义。
The vinyl acetate(VAC) polymerization rate is an important quality parameters in the process of the Poly Vinyl Alcohol production, it has significant influence on security, economy of production process and product quality. At present, it mainly relies on manual sampling off-line detection and is controlled by the operators’experiences. This paper mainly studies two problems: on-line estimation of VAC polymerization rate by using soft-sensing technique and optimizing control of VAC polymerization rate by using the method of predictive control. The concrete research works are as follows:
1. On the basis of comprehension of VAC polymerization process mechanics, various factors which affect VAC polymerization rate are analyzed and the soft-sensing model of VAC polymerization rate is established by using three methods including RBF neural network, Least Squares Support Vector Machine(LS-SVM) and Incremental Least Squares Support Vector Machines (ILS-SVM). Then the simulation based on the practical industrial data is researched and compared. The simulation results show that ILS-SVM can adjust training aims according to the error requirement and possesses high fitting and prediction precision, it has good abilities of generalization and the function of studying on line and is fit for on-line estimation of VAC polymerization rate.
2. The process of VAC polymerization has the characteristics such as nonlinearity, time-varying, disturbance and time-delay, the traditional control strategy can’t acquire satisfactory result, so the implicit generalized predictive control(GPC) algorithm is applied in the control of polymerization rate. This algorithm doesn’t depend on exact model, and its ability of disturbance rejection and time-varying restraining has been enhanced greatly with the multi-step forecasting concept. So, the implicit GPC algorithm is adopted to design for the control system of VAC polymerization rate and simulation research is done with MATLAB. The simulation results show that the implicit GPC algorithm can obtain good control performance in both the dynamic response and tracking performance. Finally, the influence of main parameters of the implicit GPC algorithm is analyzed through simulation.
Results shows that it is of importance to online prediction and optimizing control of VAC polymerization rate.
引文
[1]马延贵,牟长荣,吴三华.聚乙烯醇生产技术[M].北京:纺织工业出版社, 1988.
[2]王豪.聚乙烯醇生产的在线检测和精馏塔先进控制[J].维纶通讯,2005,Vol.25(1):38-40.
[3]邓秀江,李娜.PVA聚合工序VAC聚合率波动的原因分析及对策[J].科学技术与工程,2006,Vol.6(15):2297-2300.
[4]江西化纤化工有限责任公司.聚乙烯醇生产工艺技术规程. 2003.8.
[5]秦庆伟.PVA平均聚合度波动因素分析及优化控制[J] .天然气化工,2002,Vol.27(4):16-19.
[6]王豪.醋酸乙烯聚合反应聚合率的在线连续检测方法[P].中国专利: CN1396456,2003-02-12
[7]谢代鹤.PVA生产中醋酸乙烯聚合率的控制[J].维纶通讯,1993, Vol.13(4):17-22.
[8]吴红卫.提高聚合率合格率技术分析[J].维纶通讯,1998, Vol.18(2):17-19.
[9]洪宝庆,吴斌. Centum-CS在PVA生产中的应用[J].石油化工自动化, 2006,Vol.(5): 65-68.
[10]杨明.醋酸乙烯溶液聚合反应的研究[D]:[硕士学位论文].广西大学,2006
[11]王曦,曹柳林.醋酸乙烯聚合反应过程数学模型的建立[J].北京化工大学学报,2003,Vol.30(5):90-93
[12]李海青,黄志尧等.软测量技术原理及应用[M].北京:化学工业出版社,2000.
[13]于静江,周春晖.过程控制中的软测量技术[J].控制理论与应用,1996,Vol.13(2):137-144.
[14] Tham MT, et al. Soft-Sensors for Process Estimation and Inferential Control[J]. IEEE Trans on Neural Networks, 1991, Vol.1(1): 3-14.
[15] Dong Dong, T. J. McAvoy. Nonlinear principal component analysis-based on principal curves and neural networks[C]. American Control Conference, 1994, Vol.29 (2): 1284-1288.
[16] Wold S, et al. The Collinearity problem in linear regression: the partial least squares approach to generalized inverses[J]. SLAT, Sci. Stat. Comput ,1984(5): 743一753.
[17]刘瑞兰.软测量技术若干问题的研究及工业应用[D]:[博士学位论文].浙江大学,2004.
[18]李勇.磨矿过程参数软测量与综合优化控制的研究[D]:[博士学位论文].大连理工大学,2006.
[19]俞金寿,刘爱伦,张克进.软测量技术及其在石油化工中的应用[M].北京:化学工业出版社,2000
[20]李红军,秦永胜,徐用愚.化工过程中的数据协调及显著误差检测[J].化工自动化及仪表,1997,Vol.24(2):25-29.
[21]魏海坤.神经网络结果设计的理论与方法[M].北京:国防工业出版社,2005.
[22]王旭东,邵惠鹤.神经元网络建模和软测量技术[J].化工自动化及仪表,1996, Vol.23 (2):28-31.
[23] Willis M J, et al. Artifficial Neural Networks in Process Estimation and control [J]. Automation, 1992, Vol. 28(6): 1181-1187.
[24]朱明星,张德龙.RBF网络基函数中心选取算法的研究[J].安徽大学学报,2000, Vol.24 (1):72-78.
[25] Pena J, Lozano J, Larranaga P. An Empirical Comparison of Four Initialization Methods for the K-Means Algorithm[J]. Pattern Recognition Letters,1999, Vol.20:1027-1040.
[26]张文君,顾行发等.基于均值标准差的K均值初始聚类中心选取算法[[J].遥感学报,2006, Vol.10 (5):715-721.
[27]闻新,周露等.MATLAB神经网络仿真与应用[M].北京:科学出版社,2003.
[28]杨辉,张肃宇,李健,柴天佑.应用软测量技术实现稀土萃取分离过程的优化控制[J].吉林大学学报.2004,Vol.34(3):427-432.
[29]杨辉,谭明皓,柴天佑.基于神经网络的多元稀土萃取组分含量的软测量[J].中国稀土学报,2003,Vol.21(4):426-430
[30] Vapnik V N, The Nature of Statistical Learning Theory[M]. New York: Springer Verlag, 1995.
[31] Suykens J A K,Vandewalle J.Least squares support vector machine classfiers [J]. Neural Processing Letters. 1999, 9: 293-300.
[32] Cortes C, Vapnik V. Support-vector networks [J].Machine Learning,1995,Vol.20 (3): 273-297.
[33]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.
[34]陈文杰.基于支持向量机的聚醋工业软测量方法研究[D]: [硕士学位论文].北京化工大学,2004.
[35] Suykens J A K. Nonlinear Modeling and Support Vector Machines. IEEE Instrumentation and Measurement Technology Conference. Budapest, Hungary, 2001, Vol.1(5):287-294.
[36] Marcelo E, Suykens J A K, Bart D M. Least Squares Support Vector Machines and Primal Space Estimation. Proceedings of the 42nd IEEE Conference on Decision and Control. Maui,Hawaii USA, 2003, Vol. 4 (12):3451-3456.
[37]阎威武,朱宏栋,邵惠鹤.基于最小二乘支持向量机的软测量建模[J].系统仿真学报,2003,Vol.15(10):1494-1496.
[38]李方方,赵英凯,颜析.基于Matlab的最小二乘支持向量机的工具箱及其应用[J].计算机应用,2006, Vol.26:358-360.
[39] SUYENS JAK. LS-SVM lab Toolbox User's Guide[EB/OL]. http://www.esat kuleuven acbe/sista/,2005-12-10
[40]李悦卿.基于神经网络和最小二乘支持向量机的软测量技术应用研究[D]:[硕士学位论文] .青岛科技大学,2007.
[41]冯瑞,宋春林,张艳珠,邵惠鹤.基于支持向量机与RBF神经网络的软测量模型比较研究[J].上海交通大学学报,2003,Vol.37:122-125.
[42] Cauwenberghs G, Tomaso P.Incremental and decremental support vector machine leraning,Advances in neural information processing system (NIPS*2000), MIT Press, Cambridge, 2001,Vol.13: 409-415
[43] Vijayakumar S, Wu S. Sequential support vector classifiers and regression, Proc. Inter. Con. On Soft Computing (SOCO'99), Genoa, Italy, 1999, 610-619.
[44] Xiao R, Wang J and Zhang F. An Approach to incremental SVM learning algorithm. Proc. The 12th International Conference on Tools with Artificial Intelligence, 2000, 268-273.
[45] Suykens J, Lukas L, Vanderwallw J. Sparse approximation using least squares support vector machine. ISCAS 2000, Geneva, Switzerland, 2000, Vo1.4(5): 757-760.
[46]潘立登,潘仰东.系统辨识与建模[M].北京:化学工业出版社,2004.
[47] Lennart Ljung. Black-box models from input-output measurements. IEEE Instrumentation and Measurement Technology Conference Budapest, Hungary, 2001, 138-146.
[48]侯媛彬,汪梅,王立琦.系统辨识及其MATLAB仿真[M].北京:科学出版社,2004.
[49] Cutler,C.R, Ramaker,B.L. Dynamic Matrix Control-A Computer Control Algorithm Automatic[C]. Proc. of Joint Automatic Control Conference,1980.
[50] Rohunai R., Mehra R.K. Model Algorithmic Conrtol(MAC), Basic Theoertieal PorPerties[J].Automatica, 1982, Vol.18(4):401-414.
[51] Clarke, D.W, Mohtadi, C., Tuffs, P.S. Generalized Predictive Control[J]. Part 1 and Part 2, Automatica, 1987, Vol.23(2): 137-160.
[52] Clarke, D.W, Mohtadi, C. Properties of Generalized Predictive Control[J]. Automatica, 1989, Vol.25(6): 859-875.
[53] Clarke, D.W. Application of Generalized Predictive Control to Industrial Processes[J]. IEEE Control Syst. Maga, 1988, Vol.8(2): 49-55.
[54]林伟,孙先波.广义预测自校正控制隐式算法及其仿真研究[J].现代电子技术,2008,Vol.(7):141-143.
[55]王伟.广义预测控制理论及其应用[M].北京:科学出版社,1998
[56]谭杰,刘圣.聚乙烯生产装置关键控制回路的预测控制[J].化工自动化及仪表,2002, Vol.29(5):26-29.
[57]阚晓旭.智能控制技术在湿法磷酸生产过程中的应用研究[D]: [硕士学位论文].浙江大学,2006.
[58]朱学莉,齐维贵,伏润.GPC隐式算法及其在供热控制中的仿真研究[J].系统仿真学报,2005, Vol.17(9):2214-2217.
[59]李国勇.智能控制及其MATLAB实现[M].北京:电子工业出版社,2005.
[60]舒迪前.预测控制系统及其应用[M].北京:机械工业出版社,1996.
[61]席裕庚.预测控制[M].北京:国防工业出版社,1993.
[2]王豪.聚乙烯醇生产的在线检测和精馏塔先进控制[J].维纶通讯,2005,Vol.25(1):38-40.
[3]邓秀江,李娜.PVA聚合工序VAC聚合率波动的原因分析及对策[J].科学技术与工程,2006,Vol.6(15):2297-2300.
[4]江西化纤化工有限责任公司.聚乙烯醇生产工艺技术规程. 2003.8.
[5]秦庆伟.PVA平均聚合度波动因素分析及优化控制[J] .天然气化工,2002,Vol.27(4):16-19.
[6]王豪.醋酸乙烯聚合反应聚合率的在线连续检测方法[P].中国专利: CN1396456,2003-02-12
[7]谢代鹤.PVA生产中醋酸乙烯聚合率的控制[J].维纶通讯,1993, Vol.13(4):17-22.
[8]吴红卫.提高聚合率合格率技术分析[J].维纶通讯,1998, Vol.18(2):17-19.
[9]洪宝庆,吴斌. Centum-CS在PVA生产中的应用[J].石油化工自动化, 2006,Vol.(5): 65-68.
[10]杨明.醋酸乙烯溶液聚合反应的研究[D]:[硕士学位论文].广西大学,2006
[11]王曦,曹柳林.醋酸乙烯聚合反应过程数学模型的建立[J].北京化工大学学报,2003,Vol.30(5):90-93
[12]李海青,黄志尧等.软测量技术原理及应用[M].北京:化学工业出版社,2000.
[13]于静江,周春晖.过程控制中的软测量技术[J].控制理论与应用,1996,Vol.13(2):137-144.
[14] Tham MT, et al. Soft-Sensors for Process Estimation and Inferential Control[J]. IEEE Trans on Neural Networks, 1991, Vol.1(1): 3-14.
[15] Dong Dong, T. J. McAvoy. Nonlinear principal component analysis-based on principal curves and neural networks[C]. American Control Conference, 1994, Vol.29 (2): 1284-1288.
[16] Wold S, et al. The Collinearity problem in linear regression: the partial least squares approach to generalized inverses[J]. SLAT, Sci. Stat. Comput ,1984(5): 743一753.
[17]刘瑞兰.软测量技术若干问题的研究及工业应用[D]:[博士学位论文].浙江大学,2004.
[18]李勇.磨矿过程参数软测量与综合优化控制的研究[D]:[博士学位论文].大连理工大学,2006.
[19]俞金寿,刘爱伦,张克进.软测量技术及其在石油化工中的应用[M].北京:化学工业出版社,2000
[20]李红军,秦永胜,徐用愚.化工过程中的数据协调及显著误差检测[J].化工自动化及仪表,1997,Vol.24(2):25-29.
[21]魏海坤.神经网络结果设计的理论与方法[M].北京:国防工业出版社,2005.
[22]王旭东,邵惠鹤.神经元网络建模和软测量技术[J].化工自动化及仪表,1996, Vol.23 (2):28-31.
[23] Willis M J, et al. Artifficial Neural Networks in Process Estimation and control [J]. Automation, 1992, Vol. 28(6): 1181-1187.
[24]朱明星,张德龙.RBF网络基函数中心选取算法的研究[J].安徽大学学报,2000, Vol.24 (1):72-78.
[25] Pena J, Lozano J, Larranaga P. An Empirical Comparison of Four Initialization Methods for the K-Means Algorithm[J]. Pattern Recognition Letters,1999, Vol.20:1027-1040.
[26]张文君,顾行发等.基于均值标准差的K均值初始聚类中心选取算法[[J].遥感学报,2006, Vol.10 (5):715-721.
[27]闻新,周露等.MATLAB神经网络仿真与应用[M].北京:科学出版社,2003.
[28]杨辉,张肃宇,李健,柴天佑.应用软测量技术实现稀土萃取分离过程的优化控制[J].吉林大学学报.2004,Vol.34(3):427-432.
[29]杨辉,谭明皓,柴天佑.基于神经网络的多元稀土萃取组分含量的软测量[J].中国稀土学报,2003,Vol.21(4):426-430
[30] Vapnik V N, The Nature of Statistical Learning Theory[M]. New York: Springer Verlag, 1995.
[31] Suykens J A K,Vandewalle J.Least squares support vector machine classfiers [J]. Neural Processing Letters. 1999, 9: 293-300.
[32] Cortes C, Vapnik V. Support-vector networks [J].Machine Learning,1995,Vol.20 (3): 273-297.
[33]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.
[34]陈文杰.基于支持向量机的聚醋工业软测量方法研究[D]: [硕士学位论文].北京化工大学,2004.
[35] Suykens J A K. Nonlinear Modeling and Support Vector Machines. IEEE Instrumentation and Measurement Technology Conference. Budapest, Hungary, 2001, Vol.1(5):287-294.
[36] Marcelo E, Suykens J A K, Bart D M. Least Squares Support Vector Machines and Primal Space Estimation. Proceedings of the 42nd IEEE Conference on Decision and Control. Maui,Hawaii USA, 2003, Vol. 4 (12):3451-3456.
[37]阎威武,朱宏栋,邵惠鹤.基于最小二乘支持向量机的软测量建模[J].系统仿真学报,2003,Vol.15(10):1494-1496.
[38]李方方,赵英凯,颜析.基于Matlab的最小二乘支持向量机的工具箱及其应用[J].计算机应用,2006, Vol.26:358-360.
[39] SUYENS JAK. LS-SVM lab Toolbox User's Guide[EB/OL]. http://www.esat kuleuven acbe/sista/,2005-12-10
[40]李悦卿.基于神经网络和最小二乘支持向量机的软测量技术应用研究[D]:[硕士学位论文] .青岛科技大学,2007.
[41]冯瑞,宋春林,张艳珠,邵惠鹤.基于支持向量机与RBF神经网络的软测量模型比较研究[J].上海交通大学学报,2003,Vol.37:122-125.
[42] Cauwenberghs G, Tomaso P.Incremental and decremental support vector machine leraning,Advances in neural information processing system (NIPS*2000), MIT Press, Cambridge, 2001,Vol.13: 409-415
[43] Vijayakumar S, Wu S. Sequential support vector classifiers and regression, Proc. Inter. Con. On Soft Computing (SOCO'99), Genoa, Italy, 1999, 610-619.
[44] Xiao R, Wang J and Zhang F. An Approach to incremental SVM learning algorithm. Proc. The 12th International Conference on Tools with Artificial Intelligence, 2000, 268-273.
[45] Suykens J, Lukas L, Vanderwallw J. Sparse approximation using least squares support vector machine. ISCAS 2000, Geneva, Switzerland, 2000, Vo1.4(5): 757-760.
[46]潘立登,潘仰东.系统辨识与建模[M].北京:化学工业出版社,2004.
[47] Lennart Ljung. Black-box models from input-output measurements. IEEE Instrumentation and Measurement Technology Conference Budapest, Hungary, 2001, 138-146.
[48]侯媛彬,汪梅,王立琦.系统辨识及其MATLAB仿真[M].北京:科学出版社,2004.
[49] Cutler,C.R, Ramaker,B.L. Dynamic Matrix Control-A Computer Control Algorithm Automatic[C]. Proc. of Joint Automatic Control Conference,1980.
[50] Rohunai R., Mehra R.K. Model Algorithmic Conrtol(MAC), Basic Theoertieal PorPerties[J].Automatica, 1982, Vol.18(4):401-414.
[51] Clarke, D.W, Mohtadi, C., Tuffs, P.S. Generalized Predictive Control[J]. Part 1 and Part 2, Automatica, 1987, Vol.23(2): 137-160.
[52] Clarke, D.W, Mohtadi, C. Properties of Generalized Predictive Control[J]. Automatica, 1989, Vol.25(6): 859-875.
[53] Clarke, D.W. Application of Generalized Predictive Control to Industrial Processes[J]. IEEE Control Syst. Maga, 1988, Vol.8(2): 49-55.
[54]林伟,孙先波.广义预测自校正控制隐式算法及其仿真研究[J].现代电子技术,2008,Vol.(7):141-143.
[55]王伟.广义预测控制理论及其应用[M].北京:科学出版社,1998
[56]谭杰,刘圣.聚乙烯生产装置关键控制回路的预测控制[J].化工自动化及仪表,2002, Vol.29(5):26-29.
[57]阚晓旭.智能控制技术在湿法磷酸生产过程中的应用研究[D]: [硕士学位论文].浙江大学,2006.
[58]朱学莉,齐维贵,伏润.GPC隐式算法及其在供热控制中的仿真研究[J].系统仿真学报,2005, Vol.17(9):2214-2217.
[59]李国勇.智能控制及其MATLAB实现[M].北京:电子工业出版社,2005.
[60]舒迪前.预测控制系统及其应用[M].北京:机械工业出版社,1996.
[61]席裕庚.预测控制[M].北京:国防工业出版社,1993.