木材干燥窑自适应解耦控制器的设计与仿真
|
摘要
随着人们生活水平和居住条件的提高，人们对木材的需求量不断增加，对木材的质量要求也越来越高。这就要求我们应加强对木材干燥技术的研究和发展。
木材干燥系统是一个复杂的多变量系统，由于系统有多个输入和多个输出，内部结构比较复杂，变量之间的耦合严重。传统的解耦控制器可以在一定程度上实现解耦控制，但是其在解耦控制器的设计和实现上比较繁琐，而且解耦精度也依赖于系统数学模型的准确性，很难达到理想的控制效果。
神经网络与自适应控制的结合，为实现木材干燥的多变量解耦提供了新的思路。
本文针对木材干燥过程中温、湿度耦合的现象，提出一种将新的基于BP神经网络的PID控制器应用于木材干燥控制系统的方案，其结构和学习算法相对简单，输入层和输出层神经元物理意义明确；它根据设定的某一控制规律，通过网络的自学习，调整PID控制器的比例、积分和微分参数，从而利用经典的PID控制算法得到相应各变量的控制量参与控制，并在该过程中实现解耦控制，而不用给定样本信号进行在线的学习。最后进行了解耦仿真，取得了很好的效果。
With the improvment of people's living standards and living conditions, the demand for the quantity and quality of wood has increased higher and higher. It will boost the research and development of wood drying automatic control techniques.
Wood drying system is a complex multivariant system. There is a complex internal structure and heavy couplings among variants because of multiinputs and multioutputs. Traditional decoupling controller has a capability of decoupling control to some degree, but its design and implement is troublesome and the decoupling precision is depends on the exactness of the the system's mathematic model. So, a good control performance is difficult to be obtained.
The hybrid of neural network and adaptive control provides a new idea for decoupling multivariants wood drying control system.
In this paper, to resolve the coupling phenomena between temperature and humidity in wood drying system, a BP neural network based PID controller is proposed and applied to wood drying system. The architecture and learning algorithm of the proposed controller is more simpler and the physical meanings of the input layer's neurons and output layer's neurons are explicit. Based on predefined control rules and self-learning, the BP network changs the scaling integral and differential parameters, therefore is able to control the variants using classical PID control algorithms and at the same time, decoupling control is implemented as well during the control procedure. The decoupling control dose not need the online learning of the given signals. At the end of the paper, a decoupling simulation experiment is conducted and it has shown effective results for controlling wood drying system.
木材干燥系统是一个复杂的多变量系统，由于系统有多个输入和多个输出，内部结构比较复杂，变量之间的耦合严重。传统的解耦控制器可以在一定程度上实现解耦控制，但是其在解耦控制器的设计和实现上比较繁琐，而且解耦精度也依赖于系统数学模型的准确性，很难达到理想的控制效果。
神经网络与自适应控制的结合，为实现木材干燥的多变量解耦提供了新的思路。
本文针对木材干燥过程中温、湿度耦合的现象，提出一种将新的基于BP神经网络的PID控制器应用于木材干燥控制系统的方案，其结构和学习算法相对简单，输入层和输出层神经元物理意义明确；它根据设定的某一控制规律，通过网络的自学习，调整PID控制器的比例、积分和微分参数，从而利用经典的PID控制算法得到相应各变量的控制量参与控制，并在该过程中实现解耦控制，而不用给定样本信号进行在线的学习。最后进行了解耦仿真，取得了很好的效果。
With the improvment of people's living standards and living conditions, the demand for the quantity and quality of wood has increased higher and higher. It will boost the research and development of wood drying automatic control techniques.
Wood drying system is a complex multivariant system. There is a complex internal structure and heavy couplings among variants because of multiinputs and multioutputs. Traditional decoupling controller has a capability of decoupling control to some degree, but its design and implement is troublesome and the decoupling precision is depends on the exactness of the the system's mathematic model. So, a good control performance is difficult to be obtained.
The hybrid of neural network and adaptive control provides a new idea for decoupling multivariants wood drying control system.
In this paper, to resolve the coupling phenomena between temperature and humidity in wood drying system, a BP neural network based PID controller is proposed and applied to wood drying system. The architecture and learning algorithm of the proposed controller is more simpler and the physical meanings of the input layer's neurons and output layer's neurons are explicit. Based on predefined control rules and self-learning, the BP network changs the scaling integral and differential parameters, therefore is able to control the variants using classical PID control algorithms and at the same time, decoupling control is implemented as well during the control procedure. The decoupling control dose not need the online learning of the given signals. At the end of the paper, a decoupling simulation experiment is conducted and it has shown effective results for controlling wood drying system.
引文
[1] 张璧光．我国木材干燥技术的创新途径与发展前景．专家分析．
[2] 张璧光．从国际动态看我国木材干燥技术发展趋势．林产工业，2000，27(3)．
[3] 谢新民．丁峰编著．自适应控制系统．清华大学出版社，?
[4] 刘亚秋．木材干燥窑神经网络控制与仿真．东北林业大学硕士论文，2001年．
[5] 陶永华．自适应控制PID控制．工业仪表与自动化装置，1997，4(2)．
[6] 蒋新华．自适应PID控制．信息与控制．1988，17(5)．
[7] srtom K J, wittenmark b. Self-Tuning Controllers Based Pole-zero Placement. Proc IEE, 1980,127(3).
[8] Tjokro S, Shah S L. Adaptive PID Control. Proceedings of the 1985 American Control Conference, 1985.
[9] 夏红，王慧，李平．PID自适应控制．信息与控制，1996，25(3)．
[10] 刘国荣，阳宪惠．模糊自适应相消PID控制器．控制与决策，1995，10(2)．
[11] 谢新民．具有专家系统的PID自适应调节器．自动化仪器与仪表，1992(4)．
[12] 谭永红．神经网络自适应PID控制及其应用．模式识别与人工智能，1993，6(1)．
[13] 焦李成．神经网络的应用与实现．西安：西安电子科技大学出版社，1993．
[14] Hunt K J et al. Neural Networks for Control Systems--A Survey. Automatica, 1992, 28(6).
[15] 王永骥，徐桂英，涂建．一类工业过程的神经元智能控制．华中理工大学学报，1996，22(12)．
[16] V. Tarvainen, A. Hanhijarvi. Novel High Temperature Pressurised Kiln and Preliminary Tests on EMC and Creep of Pine and Spruce in HT Drying.
[17] Antti Hanhijarvi, Veikko Tarvainen. Benefits of computational simuliation of timber drying-improved drying ractise and understandding of drying phenomema.
[18] Zhao M. Neural-net-Based Adaptive PID Regulstor with Attenuating Excitation Signal. 1994, Proc. ACC.
[19] 王宁，泖健，陈锦江．自适应神经元网络的智能控制．信息与控制，1992，21(4)
[20] 牛玉刚，杨成梧．基于PID型神经网络的自适应控制．兵工学报，2001，22(1)
[21] Saerens M, Soquet A. A neural controller. In: proc 1st IEE Conf Neural Networts, 1989.
[22] 王东，周东华，金以慧．解耦问题的发展．自动化博览．
[23] 舒怀林．PID神经元网络对强耦合带时延多变量系统的解耦控制．控制理论与应用，1998，15(6)．
[24] 柴天佑．多变量自适应解耦控制与应用．第1版．北京：科学出版社，2001．
[25] B. Wittmark, R. Middleton and G. Goodwin, Adaptive decoupling of multivariable systems. Int. J. Control, 1987, Vol. 46, No. 6.
[26] P. E. McDermott and D. A. Mellichamp, A decoupling pole-placecent self-tuning controller for a class of multivariable process, Optimal Control Application and Method, 1986, Vol. 7.
[27] Shi-jun Lang, Xing-yun Gu and Tian-you Chai. A. multivariable generalized self-tuning controller with decoupling design. IEE Trans Autom Control, 1986, Vol. 31, No. 5.
[28] Tian-you Chai, Shi-jun Lang and Xing-yun Gu.A generalized self-tuning feedforward controller and multivariable application, Priceeding of the 24th IEEE Conference on Decision and Control, 1985.
[29] 柴天佑，郎世俊，顾兴源．多变量自校正前馈控制器及其应用．自动化学报，1996，Vol 12，No．3．
[30] 柴天佑．多变量自校正解耦控制器的全局收敛性分析．自动化学报，1989，Vol，5．
[31] 柴天佑．具有参数强一致收敛的多变量自校正前馈控制器．自动化学报，1991，Vol．17．No．3．
[32] Tian Y. Chai, A self-tuning decoupling for a class of multivariable systems and global convergence analysis, IEEE. Trans. Autom. Control. 1988 Vol. 33, No. 8.
[33] Tian You Chai, Direct adaptive decoupling control for general stochastic multivariable systems, Int. J. Control, 1990, Vol. 51, No. 4.
[34] H. Nijmeijer, Local(dynamic)input-output decoupling of discrete time nonlinear systems, Int. J. Control, 1986, Vol. 43, No. 5.
[35] 吴黎明，柴天佑．一类非线性离散时间系统的神经解耦策略．自动化学报．1997．Vol．23．No．2．
[36] A. V. Ooyen, and B. Nienhuis, Improving the convergence of the back propagation algorithm, Neural Networks, 1992, Vol. 20.
[37] 朱政贤．木材干燥．第2版．北京．中国林业出版社，1989．
[38] 梁世镇．木材干燥．第2版．北京．中国林业出版社，1986．
[39] 刘应安．木材干燥应力数学模型．东北林业大学学报，1998，26(5)．
[40] 吴跃峰．炉气间接加热木材干燥窑微机控制系统．湖南林业科技，1998，25(4)．
[41] 张乃尧，阎平凡．神经网络与模糊控制．第1版．北京．清华大学出版社，1998．
[42] 严平，钱尚源．木材干燥机理和应用研究．木工机械与木工设备，2002，30(6)．
[43] Myamoto H, Kawato M. Feedback -error-learning neural network for trajectory control of a robotic manipulator. Neural Networks, 1998, (1).
[2] 张璧光．从国际动态看我国木材干燥技术发展趋势．林产工业，2000，27(3)．
[3] 谢新民．丁峰编著．自适应控制系统．清华大学出版社，?
[4] 刘亚秋．木材干燥窑神经网络控制与仿真．东北林业大学硕士论文，2001年．
[5] 陶永华．自适应控制PID控制．工业仪表与自动化装置，1997，4(2)．
[6] 蒋新华．自适应PID控制．信息与控制．1988，17(5)．
[7] srtom K J, wittenmark b. Self-Tuning Controllers Based Pole-zero Placement. Proc IEE, 1980,127(3).
[8] Tjokro S, Shah S L. Adaptive PID Control. Proceedings of the 1985 American Control Conference, 1985.
[9] 夏红，王慧，李平．PID自适应控制．信息与控制，1996，25(3)．
[10] 刘国荣，阳宪惠．模糊自适应相消PID控制器．控制与决策，1995，10(2)．
[11] 谢新民．具有专家系统的PID自适应调节器．自动化仪器与仪表，1992(4)．
[12] 谭永红．神经网络自适应PID控制及其应用．模式识别与人工智能，1993，6(1)．
[13] 焦李成．神经网络的应用与实现．西安：西安电子科技大学出版社，1993．
[14] Hunt K J et al. Neural Networks for Control Systems--A Survey. Automatica, 1992, 28(6).
[15] 王永骥，徐桂英，涂建．一类工业过程的神经元智能控制．华中理工大学学报，1996，22(12)．
[16] V. Tarvainen, A. Hanhijarvi. Novel High Temperature Pressurised Kiln and Preliminary Tests on EMC and Creep of Pine and Spruce in HT Drying.
[17] Antti Hanhijarvi, Veikko Tarvainen. Benefits of computational simuliation of timber drying-improved drying ractise and understandding of drying phenomema.
[18] Zhao M. Neural-net-Based Adaptive PID Regulstor with Attenuating Excitation Signal. 1994, Proc. ACC.
[19] 王宁，泖健，陈锦江．自适应神经元网络的智能控制．信息与控制，1992，21(4)
[20] 牛玉刚，杨成梧．基于PID型神经网络的自适应控制．兵工学报，2001，22(1)
[21] Saerens M, Soquet A. A neural controller. In: proc 1st IEE Conf Neural Networts, 1989.
[22] 王东，周东华，金以慧．解耦问题的发展．自动化博览．
[23] 舒怀林．PID神经元网络对强耦合带时延多变量系统的解耦控制．控制理论与应用，1998，15(6)．
[24] 柴天佑．多变量自适应解耦控制与应用．第1版．北京：科学出版社，2001．
[25] B. Wittmark, R. Middleton and G. Goodwin, Adaptive decoupling of multivariable systems. Int. J. Control, 1987, Vol. 46, No. 6.
[26] P. E. McDermott and D. A. Mellichamp, A decoupling pole-placecent self-tuning controller for a class of multivariable process, Optimal Control Application and Method, 1986, Vol. 7.
[27] Shi-jun Lang, Xing-yun Gu and Tian-you Chai. A. multivariable generalized self-tuning controller with decoupling design. IEE Trans Autom Control, 1986, Vol. 31, No. 5.
[28] Tian-you Chai, Shi-jun Lang and Xing-yun Gu.A generalized self-tuning feedforward controller and multivariable application, Priceeding of the 24th IEEE Conference on Decision and Control, 1985.
[29] 柴天佑，郎世俊，顾兴源．多变量自校正前馈控制器及其应用．自动化学报，1996，Vol 12，No．3．
[30] 柴天佑．多变量自校正解耦控制器的全局收敛性分析．自动化学报，1989，Vol，5．
[31] 柴天佑．具有参数强一致收敛的多变量自校正前馈控制器．自动化学报，1991，Vol．17．No．3．
[32] Tian Y. Chai, A self-tuning decoupling for a class of multivariable systems and global convergence analysis, IEEE. Trans. Autom. Control. 1988 Vol. 33, No. 8.
[33] Tian You Chai, Direct adaptive decoupling control for general stochastic multivariable systems, Int. J. Control, 1990, Vol. 51, No. 4.
[34] H. Nijmeijer, Local(dynamic)input-output decoupling of discrete time nonlinear systems, Int. J. Control, 1986, Vol. 43, No. 5.
[35] 吴黎明，柴天佑．一类非线性离散时间系统的神经解耦策略．自动化学报．1997．Vol．23．No．2．
[36] A. V. Ooyen, and B. Nienhuis, Improving the convergence of the back propagation algorithm, Neural Networks, 1992, Vol. 20.
[37] 朱政贤．木材干燥．第2版．北京．中国林业出版社，1989．
[38] 梁世镇．木材干燥．第2版．北京．中国林业出版社，1986．
[39] 刘应安．木材干燥应力数学模型．东北林业大学学报，1998，26(5)．
[40] 吴跃峰．炉气间接加热木材干燥窑微机控制系统．湖南林业科技，1998，25(4)．
[41] 张乃尧，阎平凡．神经网络与模糊控制．第1版．北京．清华大学出版社，1998．
[42] 严平，钱尚源．木材干燥机理和应用研究．木工机械与木工设备，2002，30(6)．
[43] Myamoto H, Kawato M. Feedback -error-learning neural network for trajectory control of a robotic manipulator. Neural Networks, 1998, (1).