神经网络渠道PID控制理论及其动态仿真研究
摘要
我国已建了许多大型远距离跨流域调水工程，像已经兴建的东深供水工程、正在规划中的南水北调工程。怎样充分发挥这些工程的作用，做到优化调度、适时适量供水?一个理想供水系统，应能根据各个地区不同时间对水的实际需求，在引水口处做到适时适量的供水。如果能建立一个需求型的渠道系统，使其处于待命状态，就能满足非预定的用水要求，这就是渠系自动化的目标。
在传统渠道PID控制中一个关键的问题就是PID参数整定，传统的方法是在获取控制对象数学模型的基础上，根据某一整定原则来确定PID参数。然而实际的渠道系统是一个非常复杂的非线性、不确定系统，难于建立精确的数学模型，而且模型参数随着渠道工作状况的改变而改变，尤其对于多渠池渠道系统，要达到比较好的控制性能，必须对每个渠池的PID控制器采用不同的控制参数，而各控制参数之间又是相互影响的，参数整定难度极大。并且即使针对某一工作状况获得了PID控制的最优参数，但由于渠道一般具有时变性，仍存在整个工作范围保持最优的问题。这就要求在PID控制中参数的整定不依赖于渠道数学模型。
本文对PID控制原理、分类以及各种新型PID控制算法作了全面、系统的论述。
在传统PID渠道控制中引入了神经元和二次型性能指标，利用神经元自适应、自学习、并行处理及较强的容错能力，实现了PID控制中参数的整定不依赖于渠道数学模型，且PID参数能根据渠道的适时信息(水位、流量)在线调整以满足适时控制的要求
将BP网络与常规PID控制结合，并引入了最优化目标函数，利用BP网络来实现PID控制器参数的在线整定，将其运用于渠系等体积控制，并利用MATLAB软件进行了数值模拟仿真，取得了一些有意义的成果和较为满意的结果。
Many large projects for water-transportation have been built, such as the project of Dong-Shen water transfers in the stage of constructing and the project of south-to-north water transfers in the stage of programming. But how can we make full use of the projects? An ideal water-supply system should supply proper amount of water at proper time at the outlet. If a demand-canal was built to keep the status of ready- supply all the time, the not pre-assigned needs of water then can be satisfied. This is the just object of canal automation.
In the traditional canal PID control a key problem is how to conform the parameter of PID.The traditional way is conforming the parameter for the fundamental at the base of getting the mathematic model of the control object. But the actual canal system is a very complex non-linear unsure system. It is very difficult to establish a accurate mathematic model. Furthermore the parameters of the model change along with the working statue. Especially for the canal system of the multieach canals, it is integrant to adopt different parameters for each canal to get good control effect. But each parameter affects each other, it is very difficult to conform each parameter. And even getting the best parameters for a work status, but the canal system is up-and down, it is difficult to getting the best parameter for all work status. So it is very necessary to conform the parameters not depending on the canal mathematics model.
In the thesis the theory class and the late-model control arithmetic is summarized in the round.
Adhibiting the nerve cell and qurdratic capability target, by the self-adapting self-studying consistent and better abiding of the nerve cell ,the thesis realizes the conforming of the parameter not depending the canal mathematics ,furthermore the parameter can self-adjust for the information of the canal(water level flux).
Combining the BP nets and the routine PID controller, and adhibiting the linear optimization target function, the thesis realizes the conforming of the parameter not depending the canal mathematics, furthermore the parameter can self-adjust for the information of the canal. Using the MATLAB the thesis simulates and acquires some significative effects.
在传统渠道PID控制中一个关键的问题就是PID参数整定，传统的方法是在获取控制对象数学模型的基础上，根据某一整定原则来确定PID参数。然而实际的渠道系统是一个非常复杂的非线性、不确定系统，难于建立精确的数学模型，而且模型参数随着渠道工作状况的改变而改变，尤其对于多渠池渠道系统，要达到比较好的控制性能，必须对每个渠池的PID控制器采用不同的控制参数，而各控制参数之间又是相互影响的，参数整定难度极大。并且即使针对某一工作状况获得了PID控制的最优参数，但由于渠道一般具有时变性，仍存在整个工作范围保持最优的问题。这就要求在PID控制中参数的整定不依赖于渠道数学模型。
本文对PID控制原理、分类以及各种新型PID控制算法作了全面、系统的论述。
在传统PID渠道控制中引入了神经元和二次型性能指标，利用神经元自适应、自学习、并行处理及较强的容错能力，实现了PID控制中参数的整定不依赖于渠道数学模型，且PID参数能根据渠道的适时信息(水位、流量)在线调整以满足适时控制的要求
将BP网络与常规PID控制结合，并引入了最优化目标函数，利用BP网络来实现PID控制器参数的在线整定，将其运用于渠系等体积控制，并利用MATLAB软件进行了数值模拟仿真，取得了一些有意义的成果和较为满意的结果。
Many large projects for water-transportation have been built, such as the project of Dong-Shen water transfers in the stage of constructing and the project of south-to-north water transfers in the stage of programming. But how can we make full use of the projects? An ideal water-supply system should supply proper amount of water at proper time at the outlet. If a demand-canal was built to keep the status of ready- supply all the time, the not pre-assigned needs of water then can be satisfied. This is the just object of canal automation.
In the traditional canal PID control a key problem is how to conform the parameter of PID.The traditional way is conforming the parameter for the fundamental at the base of getting the mathematic model of the control object. But the actual canal system is a very complex non-linear unsure system. It is very difficult to establish a accurate mathematic model. Furthermore the parameters of the model change along with the working statue. Especially for the canal system of the multieach canals, it is integrant to adopt different parameters for each canal to get good control effect. But each parameter affects each other, it is very difficult to conform each parameter. And even getting the best parameters for a work status, but the canal system is up-and down, it is difficult to getting the best parameter for all work status. So it is very necessary to conform the parameters not depending on the canal mathematics model.
In the thesis the theory class and the late-model control arithmetic is summarized in the round.
Adhibiting the nerve cell and qurdratic capability target, by the self-adapting self-studying consistent and better abiding of the nerve cell ,the thesis realizes the conforming of the parameter not depending the canal mathematics ,furthermore the parameter can self-adjust for the information of the canal(water level flux).
Combining the BP nets and the routine PID controller, and adhibiting the linear optimization target function, the thesis realizes the conforming of the parameter not depending the canal mathematics, furthermore the parameter can self-adjust for the information of the canal. Using the MATLAB the thesis simulates and acquires some significative effects.
引文
[1]陶永华．新犁PID控制及其应用[M]．北京：机械工业出版社，2002，43，125
[2]郭敬枢，庄继东，孔峰，微机控制技术 重庆：重庆大学出版社，1994
[3]俞忠原，陈一民，工业过程控制计算机系统 北京：北京理工大学出版社，1995
[4]黄国建，虞平良，曾芬芳，微型计算机应用技术 上海：上海交通大学出版社，1995
[5]张宇河，金钰，计算机控制系统 北京：北京理工大学出版社，1996
[6]王耀南，基于神经网络的非线性最优控制[J]，湖南大学学报，Vol．22，No．5，Oct．1995，68-74．
[7]王耀南．智能控制系统[M]．长沙：湖南大学出版社，1996
[8]渠道运行模糊控制理论及其动态仿真研究[M]，武汉水利电力大学硕士论文，2003，2．
[9]阮新建，渠道现代挖制数学模型研究[D]，武汉大学博士沦文，2002．5．
[10]王长德，阮新建，南水北调中线总干渠控制运行设计[J]．人民长江，1999．1．
[11]阮新建，杨芳，王长德，渠道运行控制数学模型及系统特性分析，灌溉排水，Vol．34，No．1，Jan．2002
[12]阮新建，带观测器的渠道控制系统设计与运行模拟仿真，灌溉排水学报，Vol．22，No．2．Apr．2003．
[13]阮新建，渠道运行GSM算法及其适用条件[J]，中国农村水利水电，Vol．5，1999．5，15～17．
[14]杨芳，阮新建，渠道运行平衡条件及控制方程A、B矩阵的计算，中国农村水利水电，No．6，Jun．2002．
[15]王涛，阮新建，基于神经网络渠道PID控制，灌溉排水学报，2003年特刊．
[16]王涛，阮新建，基于单神经元渠道PID控制，长江科学院院报，2004．7．
[17]李宗健，江亦贞，王长德，水力自动闸门[M]，北京：水利电力出版社，1987．8，179-180．
[18]阮新建，渠道运行GSM算法及其适用条件[J]，中国农村水利水电，Vol．5，1999．5，15-17，46．
[19]孙增淅，智能控制理论与技术[M]，清华大学出版社，1997．
[20]柯善青，供水渠道等体积运行的动态过程及其数学模型的研究[M]，武汉水利电力大学硕士学位论文，2000．3．
[21]罗本成，渠系自动化运行监测系统的研究与开发[M]，武汉水利电力大学硕士学位论文，2000．4．
[22]郑邦民，槐文信，齐鄂荣，洪水水力学[M]，武汉：湖北科技出版社，2000．5，59-98．
[23]徐正凡，水力学(上册)[M]，高等教育出版社，1986．10，393-403．
[24]徐正凡，水力学(下册)[M]，高等教育出版社，1986．10，73-95，179-187，255-343．
[25]王长德，灌溉渠道水力自动控制闸门运行的稳定问题[J]，武汉水利电力学院学报，1982．4．
[26]王长德，韦直林，张礼卫，上游常水位自动控制渠道明渠非恒定流动态边界条件[J]，水利学报，No．2，1995．2，46-51，57．
[27]王长德，张礼卫，下游常水位水力自动控制渠道运行动态过程及数学模型的研究[J]，水利学报，No．11，1997．11．
[28]王长德，柯善青，冯晓波，P+PR控制器用于比威尔算法[J]，武汉水利电力大学学报，Vol．33，No．2，Apr．2000，11-15．
[29]李明思，朱凤书，渠系调控的过渡过程[J]，灌溉排水，V01．17，1998．3，49—54．
[30]J．Lewin(英国)，水工闸门的控制[J]，水利水电快报，No．2，1996，15-17，26．
[31]楼顺天，施阳，基于MATLAB的系统分析与设计——神经网络[M]，西安电子科技大学出版社，2000．
[22]薛定宇，反馈控制系统设计与分析——MATLAB语言应用[M]，清华大学出版社，2000．4．
[33]李士勇，模糊控制·神经控制和智能控制论[M]，哈尔滨工业大学出版社，1996，10．
[34]孙增淅，智能控制理论与技术[M]，清华大学出版社，1997．
canals modeled by Saint Venant's equations", to be submitted to ECC99.
[37]Xu C-Z, G.Saller, "Proportional and Integral Regulation of Irrigation canal Systens governed by the Saint Venant's equation", IEEE SMC 1998,in press
[38]Bounit H,H.Hammouri,J.Sau, "Regulation of an irrigation canal through the semigroup approach ", international Workshop on the Regulation of Irrigation Canals; (?)cte of the Art of Research and Application ,RIC97, Marrakech(Morocco),April 22-24,1997,pp 261-267
[39]Baume J-P,J.Sau, "Modeling of irrigation channel dynamic for controller design", IEEE,SMC98,in press
[40]Baume J-P, JCSau , "Study of irrigation canal dynamic for control purpose", international Workshop on the Regulation of Irrigation Canals; State of the Art of Research and Application ,RIC97, Marrakech(Morocco),April 22-24,1997, pp 261-267
[41]Liu F,J.Feyen, J.Berlamont, "Downstrem control algorithm for irrigation canals", Journal of irrigation and drainage engeneering,Vol.3,1994,pp 468-483
[42]Kosuth P, "Techniques de regulation automatique des systemes complexes ;application aux systemes hydrauliques a surface libre" These de Doctorat ,Institut National Polytechnique de Toulouse CEMAGREF LASS CNRS, 1994
[43]Litrico X, P-O.Malaterre ,D. Georges ,J.-L.Trouvat, "Commande Optimale d'un Systeme Barrage-Riviore avec Points de Mesure Intermediaires" ,Colloque SHF,25 Journees de Hydraulique, 15-17 Septembre 1998,Chambery France
[44]Fubo Liu, Jan Feyen, Pierre-Olivier Malaterre, Jean-Pierre Brume and Pascal Kosuth, Development And Evaluation of Canal Automation Algorithm CLIS [J], Journal of Irrigation and Drainage Engineering, Vol. 124, No. 1, Jan/Feb, 1998,40-46.
[45]Schuurmans J, "control of water levels in open channels" ,PHD thesis ,14 October 1997,223 p, ISBN 90-90[0995-],delft,Netherland
[46]Schuurmans J,O.H.Bosgra,R.Brouwer, "Open-channel flow model approximation for controller design" ,TU Delft,Appl,Math Modelling 1995,Vol. 19.September.pp.525-530
[47]Sanfilippo F, "Application de I'automatique multivariable a la regulation des canaux
Comoaraison avec une approche monovariable " ,24 Octobre 1997,These de Doctoral ,Universite Claude Bemard,Lyon 1,scp,France
[48]Toudeft A, "Neural control of nonlinear non-minimum phase dynamic systems " ,in the Proco of the Intern Conc on Artificial Neual Nets and Genetic Algorithms ,Alex,France,May, 18-21,1995 .
[49]Toudeft A, "Combining neural and adaptive controllers for a non-minimum phase varying time delay system", Journees Hispano-Francaises : Systemes intelligents and controle avance, Barcelona(Spain),Universite,Polytechnique, de Catalogne , November 12-13,1996
[50]Stringam B.L, G.P. Merkley, "Field Application of a fuzzy controller for an irrigation canal in Roosevelt Utah" , international Workshop on the Regulation of Irrigation Canals; State of the Art of Research and Application , RIC97, Marrakech(Morocco),April 22-24,1997,pp 349-354
[51]Stringam B.L, G.P. Merkley, "Fuzzy controller simulation for local downstream water level control in canals", international Workshop on the Regulation of Irrigation Canals; State of the Art of Research and Application , RIC97, Marrakech(Morocco),April 22-24,1997,pp 342-348
[52]Voron B, A.P. Bouillot, "Application of the fuzzy set theory to the control of a large canal", international Workshop on the Regulation of Irrigation Canals; State of the Art of Research and Application ,RIC97, Marrakech(Morocco),April 22-24,1997,pp 317-341
[2]郭敬枢，庄继东，孔峰，微机控制技术 重庆：重庆大学出版社，1994
[3]俞忠原，陈一民，工业过程控制计算机系统 北京：北京理工大学出版社，1995
[4]黄国建，虞平良，曾芬芳，微型计算机应用技术 上海：上海交通大学出版社，1995
[5]张宇河，金钰，计算机控制系统 北京：北京理工大学出版社，1996
[6]王耀南，基于神经网络的非线性最优控制[J]，湖南大学学报，Vol．22，No．5，Oct．1995，68-74．
[7]王耀南．智能控制系统[M]．长沙：湖南大学出版社，1996
[8]渠道运行模糊控制理论及其动态仿真研究[M]，武汉水利电力大学硕士论文，2003，2．
[9]阮新建，渠道现代挖制数学模型研究[D]，武汉大学博士沦文，2002．5．
[10]王长德，阮新建，南水北调中线总干渠控制运行设计[J]．人民长江，1999．1．
[11]阮新建，杨芳，王长德，渠道运行控制数学模型及系统特性分析，灌溉排水，Vol．34，No．1，Jan．2002
[12]阮新建，带观测器的渠道控制系统设计与运行模拟仿真，灌溉排水学报，Vol．22，No．2．Apr．2003．
[13]阮新建，渠道运行GSM算法及其适用条件[J]，中国农村水利水电，Vol．5，1999．5，15～17．
[14]杨芳，阮新建，渠道运行平衡条件及控制方程A、B矩阵的计算，中国农村水利水电，No．6，Jun．2002．
[15]王涛，阮新建，基于神经网络渠道PID控制，灌溉排水学报，2003年特刊．
[16]王涛，阮新建，基于单神经元渠道PID控制，长江科学院院报，2004．7．
[17]李宗健，江亦贞，王长德，水力自动闸门[M]，北京：水利电力出版社，1987．8，179-180．
[18]阮新建，渠道运行GSM算法及其适用条件[J]，中国农村水利水电，Vol．5，1999．5，15-17，46．
[19]孙增淅，智能控制理论与技术[M]，清华大学出版社，1997．
[20]柯善青，供水渠道等体积运行的动态过程及其数学模型的研究[M]，武汉水利电力大学硕士学位论文，2000．3．
[21]罗本成，渠系自动化运行监测系统的研究与开发[M]，武汉水利电力大学硕士学位论文，2000．4．
[22]郑邦民，槐文信，齐鄂荣，洪水水力学[M]，武汉：湖北科技出版社，2000．5，59-98．
[23]徐正凡，水力学(上册)[M]，高等教育出版社，1986．10，393-403．
[24]徐正凡，水力学(下册)[M]，高等教育出版社，1986．10，73-95，179-187，255-343．
[25]王长德，灌溉渠道水力自动控制闸门运行的稳定问题[J]，武汉水利电力学院学报，1982．4．
[26]王长德，韦直林，张礼卫，上游常水位自动控制渠道明渠非恒定流动态边界条件[J]，水利学报，No．2，1995．2，46-51，57．
[27]王长德，张礼卫，下游常水位水力自动控制渠道运行动态过程及数学模型的研究[J]，水利学报，No．11，1997．11．
[28]王长德，柯善青，冯晓波，P+PR控制器用于比威尔算法[J]，武汉水利电力大学学报，Vol．33，No．2，Apr．2000，11-15．
[29]李明思，朱凤书，渠系调控的过渡过程[J]，灌溉排水，V01．17，1998．3，49—54．
[30]J．Lewin(英国)，水工闸门的控制[J]，水利水电快报，No．2，1996，15-17，26．
[31]楼顺天，施阳，基于MATLAB的系统分析与设计——神经网络[M]，西安电子科技大学出版社，2000．
[22]薛定宇，反馈控制系统设计与分析——MATLAB语言应用[M]，清华大学出版社，2000．4．
[33]李士勇，模糊控制·神经控制和智能控制论[M]，哈尔滨工业大学出版社，1996，10．
[34]孙增淅，智能控制理论与技术[M]，清华大学出版社，1997．
canals modeled by Saint Venant's equations", to be submitted to ECC99.
[37]Xu C-Z, G.Saller, "Proportional and Integral Regulation of Irrigation canal Systens governed by the Saint Venant's equation", IEEE SMC 1998,in press
[38]Bounit H,H.Hammouri,J.Sau, "Regulation of an irrigation canal through the semigroup approach ", international Workshop on the Regulation of Irrigation Canals; (?)cte of the Art of Research and Application ,RIC97, Marrakech(Morocco),April 22-24,1997,pp 261-267
[39]Baume J-P,J.Sau, "Modeling of irrigation channel dynamic for controller design", IEEE,SMC98,in press
[40]Baume J-P, JCSau , "Study of irrigation canal dynamic for control purpose", international Workshop on the Regulation of Irrigation Canals; State of the Art of Research and Application ,RIC97, Marrakech(Morocco),April 22-24,1997, pp 261-267
[41]Liu F,J.Feyen, J.Berlamont, "Downstrem control algorithm for irrigation canals", Journal of irrigation and drainage engeneering,Vol.3,1994,pp 468-483
[42]Kosuth P, "Techniques de regulation automatique des systemes complexes ;application aux systemes hydrauliques a surface libre" These de Doctorat ,Institut National Polytechnique de Toulouse CEMAGREF LASS CNRS, 1994
[43]Litrico X, P-O.Malaterre ,D. Georges ,J.-L.Trouvat, "Commande Optimale d'un Systeme Barrage-Riviore avec Points de Mesure Intermediaires" ,Colloque SHF,25 Journees de Hydraulique, 15-17 Septembre 1998,Chambery France
[44]Fubo Liu, Jan Feyen, Pierre-Olivier Malaterre, Jean-Pierre Brume and Pascal Kosuth, Development And Evaluation of Canal Automation Algorithm CLIS [J], Journal of Irrigation and Drainage Engineering, Vol. 124, No. 1, Jan/Feb, 1998,40-46.
[45]Schuurmans J, "control of water levels in open channels" ,PHD thesis ,14 October 1997,223 p, ISBN 90-90[0995-],delft,Netherland
[46]Schuurmans J,O.H.Bosgra,R.Brouwer, "Open-channel flow model approximation for controller design" ,TU Delft,Appl,Math Modelling 1995,Vol. 19.September.pp.525-530
[47]Sanfilippo F, "Application de I'automatique multivariable a la regulation des canaux
Comoaraison avec une approche monovariable " ,24 Octobre 1997,These de Doctoral ,Universite Claude Bemard,Lyon 1,scp,France
[48]Toudeft A, "Neural control of nonlinear non-minimum phase dynamic systems " ,in the Proco of the Intern Conc on Artificial Neual Nets and Genetic Algorithms ,Alex,France,May, 18-21,1995 .
[49]Toudeft A, "Combining neural and adaptive controllers for a non-minimum phase varying time delay system", Journees Hispano-Francaises : Systemes intelligents and controle avance, Barcelona(Spain),Universite,Polytechnique, de Catalogne , November 12-13,1996
[50]Stringam B.L, G.P. Merkley, "Field Application of a fuzzy controller for an irrigation canal in Roosevelt Utah" , international Workshop on the Regulation of Irrigation Canals; State of the Art of Research and Application , RIC97, Marrakech(Morocco),April 22-24,1997,pp 349-354
[51]Stringam B.L, G.P. Merkley, "Fuzzy controller simulation for local downstream water level control in canals", international Workshop on the Regulation of Irrigation Canals; State of the Art of Research and Application , RIC97, Marrakech(Morocco),April 22-24,1997,pp 342-348
[52]Voron B, A.P. Bouillot, "Application of the fuzzy set theory to the control of a large canal", international Workshop on the Regulation of Irrigation Canals; State of the Art of Research and Application ,RIC97, Marrakech(Morocco),April 22-24,1997,pp 317-341