基于模块化思想的CGPC控制软件的开发和应用
摘要
传统的预测控制算法为了适合计算机实现,所使用的预测模型都是基于离散时间的,即使辨识出连续时间模型,也要在合理的采样时间选择下做离散化处理,以适应基于离散时间模型的预测控制算法的需要。这样在模型离散化的过程中,受到采样时间及近似方法的局限,难免会产生模型误差。通常这个采样时间的选择主要靠长期的经验和扎实的控制理论知识,如果选择不当,甚至会导致系统不稳定。尤其是复杂的多变量系统包含有许多具有不同时间常数的过程,若为其设计离散时间的多变量控制系统,就不得不采用多采样速率。在实际生产中,很多化工企业购买了先进控制软件包,这些先进控制软件包在投入使用初期效果明显,但能长期投入运行的却是寥寥无几。分析其原因,这些先进控制软件包内部结构复杂,随着装置的运行,模型发生变化,而先进控制软件包所使用的模型不能在线及时更新或难于更新,结果不得不放弃这些先进控制软件包。
本文针对这些控制软件包的缺点引入连续时间广义预测控制(CGPC),它的适用范围更广、与时间连续的真实世界更加贴切,避免了离散时间GPC方法中存在的诸如数值灵敏度、采样速率选择、非最小相位零点等问题,而且充分发挥了连续时间方法的优点,具有良好的鲁棒性能和抗干扰能力。为此从实际应用角度出发,开发了连续时间广义预测控制软件,该软件采用OPC数据采集技术,以基于随机数搜索(NLJ)的连续时间模型辨识方法为基础,实现了连续时间控制律的模块化仿真与在线控制。通过该软件平台可以计算具有多种组合形式的连续时间控制律,并能实时在线的观察各个模块的输入、输出以及状态,设计整定控制器参数。该软件开发过程中引入模块化思想,将控制律分解为不同的信号流模块,单独计算,能够充分发挥连续广义预测控制器中观测器的重要作用,解决了复杂控制律的实现问题,为在线控制奠定基础。通过对济南炼油厂一个实际模型的控制效果,证明了该软件的有效性,推广了连续时间控制律的应用。
For the convenience of application with computer, traditional module predictive control (MPC) arithmetic is based on discrete-time model. Even the identified continuous-time models have to be discretized in order to meet the need of MPC arithmetic which is on the ground of discrete model. On account of the restriction of sample time and approximate method, model error is unavailable in the discretizing of model. Usually the choice of sample time is found on long time experiences and well-knit control theory, if it is chose improper, the system will be unstable. Especially, in order to design the discrete multivariable control system for the complicated multivariable system with many processes which have different time constant, a lot of sample time will have to be brought in. In the practical process, many chemical plants bought advanced control software which usually performs well in the beginning but can hardly last for long time. The reasons are as follows: the inner structure of these programs is too complicated to update in real-time for the shift of model, and they have to be discarded in the end. In allusion to problems above, this essay brings in continuous-time predictive control (CGPC) theory, which has wider applicability and more suitable for the real process, avoids the sensitivity of data, the choice of sample time and zeros of non-minimum phase in discrete GPC, moreover CGPC exerts the advantages of continuous-time theory and performs robustly and stably. So in a practical view, a continuous-time generalized predictive control program is developed, which is on the ground of NLJ arithmetic model identification theory and realizes the simulation and online control of continuous-time control law. It is available to calculate many kinds of continuous-time control law, access and observe the input, output and status of every module, design and set the control parameters. In the development of this program, module theory was brought in. The control law was decomposed into different signal flow modules which were calculated separately so as to make full use of observer in CGPC controller, realize the complicated control laws and in the end establish the fundamental of online control. Through the successful application on a model in Jinan refinery, this software is proved to be effective and extends the application of continuous-time control laws.
本文针对这些控制软件包的缺点引入连续时间广义预测控制(CGPC),它的适用范围更广、与时间连续的真实世界更加贴切,避免了离散时间GPC方法中存在的诸如数值灵敏度、采样速率选择、非最小相位零点等问题,而且充分发挥了连续时间方法的优点,具有良好的鲁棒性能和抗干扰能力。为此从实际应用角度出发,开发了连续时间广义预测控制软件,该软件采用OPC数据采集技术,以基于随机数搜索(NLJ)的连续时间模型辨识方法为基础,实现了连续时间控制律的模块化仿真与在线控制。通过该软件平台可以计算具有多种组合形式的连续时间控制律,并能实时在线的观察各个模块的输入、输出以及状态,设计整定控制器参数。该软件开发过程中引入模块化思想,将控制律分解为不同的信号流模块,单独计算,能够充分发挥连续广义预测控制器中观测器的重要作用,解决了复杂控制律的实现问题,为在线控制奠定基础。通过对济南炼油厂一个实际模型的控制效果,证明了该软件的有效性,推广了连续时间控制律的应用。
For the convenience of application with computer, traditional module predictive control (MPC) arithmetic is based on discrete-time model. Even the identified continuous-time models have to be discretized in order to meet the need of MPC arithmetic which is on the ground of discrete model. On account of the restriction of sample time and approximate method, model error is unavailable in the discretizing of model. Usually the choice of sample time is found on long time experiences and well-knit control theory, if it is chose improper, the system will be unstable. Especially, in order to design the discrete multivariable control system for the complicated multivariable system with many processes which have different time constant, a lot of sample time will have to be brought in. In the practical process, many chemical plants bought advanced control software which usually performs well in the beginning but can hardly last for long time. The reasons are as follows: the inner structure of these programs is too complicated to update in real-time for the shift of model, and they have to be discarded in the end. In allusion to problems above, this essay brings in continuous-time predictive control (CGPC) theory, which has wider applicability and more suitable for the real process, avoids the sensitivity of data, the choice of sample time and zeros of non-minimum phase in discrete GPC, moreover CGPC exerts the advantages of continuous-time theory and performs robustly and stably. So in a practical view, a continuous-time generalized predictive control program is developed, which is on the ground of NLJ arithmetic model identification theory and realizes the simulation and online control of continuous-time control law. It is available to calculate many kinds of continuous-time control law, access and observe the input, output and status of every module, design and set the control parameters. In the development of this program, module theory was brought in. The control law was decomposed into different signal flow modules which were calculated separately so as to make full use of observer in CGPC controller, realize the complicated control laws and in the end establish the fundamental of online control. Through the successful application on a model in Jinan refinery, this software is proved to be effective and extends the application of continuous-time control laws.
引文
[1]席裕庚,预测控制[M],第一版,北京,国防工业出版社,1993
[2]王树青,先进控制技术及应用[M],第一版,北京,化工出版社,2001
[3]Qin S J,Badgwell T A.An survey of industrial model predictive control technology[J].Control Engineering Practice,2003,11(7):733-764
[4]Gattu G,Zafirou E.Observer Based Nonlinear Quadratic Dynamic Matrix Control for State Space and Input/Output Models[J].Canadian Journal of Chemical Engineering,1995,73(6):883-895
[5]Culter,C.R.,Ramaker,B.L.Dynamic matrix control-a computer control algorithm[C].In Proceeding of the joint automatic control conference,1980
[6]Culter,C.R.,Morshedi,A.,Haydel,J.An industrial perspective on advanced control[C].In RICHE annual meeting,Washington,DC,October,1983
[7]Garcia,C.E,Morshedi,A.M.Quadratic programming solution of dynamic matrix control (QDMC)[J].Chemical Engineering Communications,1986,46:73-87
[8]Embirucu M,Fontes C.Multirate multivariable generalized predictive control and its application to a slurry reactor for ethylene polymerization[J].Chemical Engineering Science,2006,61(17):5754-5767
[9]DMC Corp.DMC Technology overview[R].Product literature from DMC Corp.,July 1994
[10]徐立鸿.预测控制的研究现状及问题[J].控制理论与应用.1994,11(1):121-125
[11]李平,王树青,王骥程.预测控制研究的概况[J].化工自动化及仪表1995.22(6):3-9
[12]Mehra,R.K.,R.Rouhani,J.Eremo,J.Richalet,Rauh,A.Model Algorithmic Control:Review and Recent Development[C]Eng.Foundation Conf.on Chemical Process Control 2,Sea Island,Georgia,1982,2(14):287-310
[13]Clarke,D.W.Generalized predictive control[J].Automatica,1987,23(2):137-160
[14]Demircoiglu H,Ercan Karasu.Generalized predictive control:A practical application and comparison ofdiscrete and continuous-time versions[J].IEEE Control Systems Magazine,2000,10(1):36-47
[15]周立峰,讲座预测控制——第一讲预测控制的基本原理[J],自动化与仪器仪表,1995,5(12):56-60.
[16]马文学,钟汉枢,动态矩阵控制研究进展及其应用现状[J],重庆工业高等专科学校学报,2005,20(1):49-54.
[17]李金霞,邱公伟,动态矩阵控制及其改进方法的仿真研究[J].福州大学学报,2004,32(5):560-562
[18]王树青,金晓明.先进控制技术应用实例[M].北京:化学工业出版社,2005
[19]邹涛,李少远.基于PC平台的预测控制软件包开发与应用[J].中南工业大学学报(自然科学版).2003,34(4):380-382
[20]赵景波,周祥龙.预测函数控制在工业锅炉燃烧系统上的应用[J].工业仪表与自动化装置,2003,(3):24-26
[21]胡家升,潘红华,苏宏业等.预测函数控制系统的闭环性能分析[J].控制理论与应用,2001,18(5):774-778
[22]邹媛媛,刘晓华,李少远.满足因果约束的多速率单值广义预测控制[J].控制与决策. 2005,20(7):827-829
[23]M.Embirucu and C.Fontes.Multirate multivariable generalized predictive control and its application to a slurry reactor for ethylene polymerization[J].Chemical Engineering Science.61(17),2006:5754-5767
[24]Demircioglu H.,Gawthrop P.J.Continuous-time Generalized Predictive Control(CGPC)[J].Automatica.1991,27(1):55-74
[25]Demiricioglu H.and Gawthrop P.J..Multivariable continuous-time generalized predictive control(MCGPC)[J].Automatica,1992,28(4):697-713
[26]Demiricioglu H.and Clarke D.W..CGPG with guaranteed stability properties[J].IEE Proc-D,1992,139(4):371-380
[27]Demiricioglu H.and Ercan Karasu.Generalized Predictive Control- A Practical Application and Comparison of Discrete-and Continuous-Time Versions[J].IEEE Control Systems Magazine,2000,10:36-47
[28]Demiricioglu H..Constrained continuous-time generalized predictive control[J].IEE Proceeding D,1999,146(5):470-476
[29]Demircoiglu H and Yavuzyilmaz.C.Constrained Predictive Control in Continuous Time [J].IEEE Control Systems Magazine,2002(8):57-67
[30]Kowalczuk Z.and Suchomski P..Continuous-time generalized predictive control of delay svstems[J].IEE Proc-Control Theorv Appl..1999.146(1):65-75
[31]姜景杰,甄新平,李全善,魏环,靳其兵,潘立登.一种基于改进NLJ算法的对象辨识方法及其应用[J].中国化学工程学报(英文版).2007,15(1):88-91
[32]李巍,王明明,潘立登矩阵连接法在辨识和pid控制器参数设计中的应用[J]。石油化工高等学校学报,2007,9(z1):9-11
[33]李明,陈宗海,张海涛,秦廷.连续时间广义预测控制算法的应用研究[J].测控技术.2004,23(7):61-63
[34]杨智,刘暾东,赵克中.一种新的模糊内模预估控制器[J].甘肃科学学报,2000,12(1):12-18
[35]赵文杰,刘吉臻,金秀章,王东风.时滞工业对象的一种建模和控制方法[J].华北电力大学学报,2001,28(4):29-32
[36]田涛,杨志远,罗毅,吕跃刚.一种Smith预估系统的鲁棒PID控制算法[J].现代电力,1998,15(2):8-13
[37]贾立,杨智,杨李成.一种新的时变时滞系统控制方法[J].甘肃工业大学学报,1999,25(4):52-56
[38]张井岗,李临生,陈志梅,赵志诚.二自由度PID调节器的内模整定方法[J].仪器仪表学报,2002,23(1):28-30
[39]袁树清,华军,许庆余,张陵.内模控制及其在机构振动主动控制中的应用[J].应用力学学报,1999,16(2):48-53
[40]孙鑫,孔亚广,孙优贤.造纸过程定量水分的建模与控制[J].中国造纸,2000,5(6):19-23
[41]郑思让,张玲.内模控制在造纸生产定量控制中的应用[J].应用力学学报,2001,28(3):23-24
[42]许天舒,史小平.核电战反应堆冷却剂平均温度内模控制系统仿真[J].哈尔滨理工大学学报,2001,26(2):26-30
[43]陈志辉,严仰光.IMC控制在双凸极发电机电压调节中的应用[J].电工技术杂志, 2000,6(10):28-29
[44]徐国君,李成利.OPC技术及其在工业控制中的应用[J].青岛科技大学学报,2003,24(6):545-547
[45]郭红晓,莫德举.OPC技术及其软件的开发[J].北京化工大学学报,2002,29(3):71-74.
[46]Rein Luus,Jaakola.Optimization by direct search and systematic reduction of the size of search region[J].AIChE,1973,19(4):760-766
[47]潘立登,聂雪媛,马俊英.常减压蒸馏装置监控先进控制和优化控制.石油化工自动化,2001,6(10):10-13
[48]潘红华,苏宏业,褚健.具有时滞和干扰的多变量系统预测函数控制方法的研究[J].控制理论与应用,2000,17(6):813-818
[49]何健辉,董方鹏.实用Visual C++6.0教程[M],北京:清华大学出版社
[50]粟塔山.VC++和BC++数值分析类库[M],北京:清华大学出版社,2005
[2]王树青,先进控制技术及应用[M],第一版,北京,化工出版社,2001
[3]Qin S J,Badgwell T A.An survey of industrial model predictive control technology[J].Control Engineering Practice,2003,11(7):733-764
[4]Gattu G,Zafirou E.Observer Based Nonlinear Quadratic Dynamic Matrix Control for State Space and Input/Output Models[J].Canadian Journal of Chemical Engineering,1995,73(6):883-895
[5]Culter,C.R.,Ramaker,B.L.Dynamic matrix control-a computer control algorithm[C].In Proceeding of the joint automatic control conference,1980
[6]Culter,C.R.,Morshedi,A.,Haydel,J.An industrial perspective on advanced control[C].In RICHE annual meeting,Washington,DC,October,1983
[7]Garcia,C.E,Morshedi,A.M.Quadratic programming solution of dynamic matrix control (QDMC)[J].Chemical Engineering Communications,1986,46:73-87
[8]Embirucu M,Fontes C.Multirate multivariable generalized predictive control and its application to a slurry reactor for ethylene polymerization[J].Chemical Engineering Science,2006,61(17):5754-5767
[9]DMC Corp.DMC Technology overview[R].Product literature from DMC Corp.,July 1994
[10]徐立鸿.预测控制的研究现状及问题[J].控制理论与应用.1994,11(1):121-125
[11]李平,王树青,王骥程.预测控制研究的概况[J].化工自动化及仪表1995.22(6):3-9
[12]Mehra,R.K.,R.Rouhani,J.Eremo,J.Richalet,Rauh,A.Model Algorithmic Control:Review and Recent Development[C]Eng.Foundation Conf.on Chemical Process Control 2,Sea Island,Georgia,1982,2(14):287-310
[13]Clarke,D.W.Generalized predictive control[J].Automatica,1987,23(2):137-160
[14]Demircoiglu H,Ercan Karasu.Generalized predictive control:A practical application and comparison ofdiscrete and continuous-time versions[J].IEEE Control Systems Magazine,2000,10(1):36-47
[15]周立峰,讲座预测控制——第一讲预测控制的基本原理[J],自动化与仪器仪表,1995,5(12):56-60.
[16]马文学,钟汉枢,动态矩阵控制研究进展及其应用现状[J],重庆工业高等专科学校学报,2005,20(1):49-54.
[17]李金霞,邱公伟,动态矩阵控制及其改进方法的仿真研究[J].福州大学学报,2004,32(5):560-562
[18]王树青,金晓明.先进控制技术应用实例[M].北京:化学工业出版社,2005
[19]邹涛,李少远.基于PC平台的预测控制软件包开发与应用[J].中南工业大学学报(自然科学版).2003,34(4):380-382
[20]赵景波,周祥龙.预测函数控制在工业锅炉燃烧系统上的应用[J].工业仪表与自动化装置,2003,(3):24-26
[21]胡家升,潘红华,苏宏业等.预测函数控制系统的闭环性能分析[J].控制理论与应用,2001,18(5):774-778
[22]邹媛媛,刘晓华,李少远.满足因果约束的多速率单值广义预测控制[J].控制与决策. 2005,20(7):827-829
[23]M.Embirucu and C.Fontes.Multirate multivariable generalized predictive control and its application to a slurry reactor for ethylene polymerization[J].Chemical Engineering Science.61(17),2006:5754-5767
[24]Demircioglu H.,Gawthrop P.J.Continuous-time Generalized Predictive Control(CGPC)[J].Automatica.1991,27(1):55-74
[25]Demiricioglu H.and Gawthrop P.J..Multivariable continuous-time generalized predictive control(MCGPC)[J].Automatica,1992,28(4):697-713
[26]Demiricioglu H.and Clarke D.W..CGPG with guaranteed stability properties[J].IEE Proc-D,1992,139(4):371-380
[27]Demiricioglu H.and Ercan Karasu.Generalized Predictive Control- A Practical Application and Comparison of Discrete-and Continuous-Time Versions[J].IEEE Control Systems Magazine,2000,10:36-47
[28]Demiricioglu H..Constrained continuous-time generalized predictive control[J].IEE Proceeding D,1999,146(5):470-476
[29]Demircoiglu H and Yavuzyilmaz.C.Constrained Predictive Control in Continuous Time [J].IEEE Control Systems Magazine,2002(8):57-67
[30]Kowalczuk Z.and Suchomski P..Continuous-time generalized predictive control of delay svstems[J].IEE Proc-Control Theorv Appl..1999.146(1):65-75
[31]姜景杰,甄新平,李全善,魏环,靳其兵,潘立登.一种基于改进NLJ算法的对象辨识方法及其应用[J].中国化学工程学报(英文版).2007,15(1):88-91
[32]李巍,王明明,潘立登矩阵连接法在辨识和pid控制器参数设计中的应用[J]。石油化工高等学校学报,2007,9(z1):9-11
[33]李明,陈宗海,张海涛,秦廷.连续时间广义预测控制算法的应用研究[J].测控技术.2004,23(7):61-63
[34]杨智,刘暾东,赵克中.一种新的模糊内模预估控制器[J].甘肃科学学报,2000,12(1):12-18
[35]赵文杰,刘吉臻,金秀章,王东风.时滞工业对象的一种建模和控制方法[J].华北电力大学学报,2001,28(4):29-32
[36]田涛,杨志远,罗毅,吕跃刚.一种Smith预估系统的鲁棒PID控制算法[J].现代电力,1998,15(2):8-13
[37]贾立,杨智,杨李成.一种新的时变时滞系统控制方法[J].甘肃工业大学学报,1999,25(4):52-56
[38]张井岗,李临生,陈志梅,赵志诚.二自由度PID调节器的内模整定方法[J].仪器仪表学报,2002,23(1):28-30
[39]袁树清,华军,许庆余,张陵.内模控制及其在机构振动主动控制中的应用[J].应用力学学报,1999,16(2):48-53
[40]孙鑫,孔亚广,孙优贤.造纸过程定量水分的建模与控制[J].中国造纸,2000,5(6):19-23
[41]郑思让,张玲.内模控制在造纸生产定量控制中的应用[J].应用力学学报,2001,28(3):23-24
[42]许天舒,史小平.核电战反应堆冷却剂平均温度内模控制系统仿真[J].哈尔滨理工大学学报,2001,26(2):26-30
[43]陈志辉,严仰光.IMC控制在双凸极发电机电压调节中的应用[J].电工技术杂志, 2000,6(10):28-29
[44]徐国君,李成利.OPC技术及其在工业控制中的应用[J].青岛科技大学学报,2003,24(6):545-547
[45]郭红晓,莫德举.OPC技术及其软件的开发[J].北京化工大学学报,2002,29(3):71-74.
[46]Rein Luus,Jaakola.Optimization by direct search and systematic reduction of the size of search region[J].AIChE,1973,19(4):760-766
[47]潘立登,聂雪媛,马俊英.常减压蒸馏装置监控先进控制和优化控制.石油化工自动化,2001,6(10):10-13
[48]潘红华,苏宏业,褚健.具有时滞和干扰的多变量系统预测函数控制方法的研究[J].控制理论与应用,2000,17(6):813-818
[49]何健辉,董方鹏.实用Visual C++6.0教程[M],北京:清华大学出版社
[50]粟塔山.VC++和BC++数值分析类库[M],北京:清华大学出版社,2005