预测自适应双重控制的研究与应用
|
摘要
预测控制是一种基于模型的先进控制算法,它通过预测被控对象的输出并结合反馈校正来决定其最优控制作用。预测控制也是目前仅有的成功应用于工业控制中的先进控制方法之一。由于实际生产过程的数学模型往往难以确知或者由于工作情况改变或环境变化造成被控对象特性的改变,对于这类系统,我们通常采用自适应控制解决问题。为了进一步改善自适应控制,用先进的自适应双重控制代替原有的控制算法。自适应双重控制是一种先进的自适应控制算法,它有其固有的优点但也存在局限性。本文利用自适应双重控制的优点,对预测控制算法进行改进,改善预测控制效果。并对自适应双重控制本身的缺陷在分析研究的基础上提出改进办法,使其得到更广泛的应用。具体研究成果如下:
1、提出了一种基于双重控制思想的广义预测自适应控制算法,该算法在模型辨识和控制的过程中,将模型的准确程度作为设计控制律的一个重要因素,在模型失配的条件下,抑制控制量的波动,同时最大限度的积累被估计参数的信息,以便最快地降低系统的不确定性。仿真结果表明,该控制算法比普通的广义预测自适应控制具有更好的控制品质,并将此方法推广到先进的预测函数控制中,效果同样令人满意。
2、提出了基于Hammerstein非线性模型的自适应双重预测控制策略。采用Hammerstein非线性模型线性化的方法,将线性部分与非线性部分有机的分离,对于线性部分作自适应双重预测控制,非线性部分当作一个求解高阶多项式的过程。通过对PH中和滴定过程的仿真实验可以看到该方法具有良好的控制品质。
3、针对系统稳态时发生大扰动或在系统变结构的情况下,自适应作用弱化,模型更新缓慢的缺点,提出了基于系统波动振幅的条件准则,强制提升双重控制作用的方案。该方案充分利用自适应双重控制所具有的良好抗干扰性能,使其抗干扰能力不仅仅在模型辨识的初始阶段发挥作用,而且在整个系统动态过程中发挥抗干扰作用。
4、针对自适应双重控制目前还存在的一些问题,比如关断现象,参数调整困难等,引入协调因子β,协调辨识得到的协方差矩阵中pb 1( k )与关键参数(b_1~2)|∧(k)的比例关系,避免由于相差悬殊而出现关断现象。另外,寻找模型中的调节参数β与模型本身的关系,确定不同被控对象下模型参数的大致范围,在一定范围内调节参数,减少工作量。
Predictive control is an advanced control algorithms based on models. It uses the forecast output and feedback correction to determine its optimal control action. Predictive control has three segments including forecasting model, rolling optimization and feedback correction. In the process of predictive computing, whether the model is matching the object is a key factor. In order to identify the parameters of the model, the adaptive control method is used. Adaptive dual control (ADC) is an advanced adaptive algorithm. It has its inherent advantages and limitations. By means of the advantages of ADC, this dissertation improves the predictive control algorithms, and reforms the limitations of the ADC, so that it can be used widely. The detail studies as follows:
1. A dual control algorithm of generalized predictive adaptive control (GPADC) is presented, it considers errors of estimated parameters sufficiently in the process of identifying model, and inhabits the fluctuation by this important factor. This algorithm accumulates the information of estimated parameters maximal at the same time. The simulation results show that this control algorithm is more effective than common generalized predictive adaptive control. Then apply adaptive dual control algorithm in predictive function control, the result is also satisfactory.
2. An adaptive dual predictive control algorithm based on nonlinear Hammerstein model is presented, using the linearization method the Hammerstein model is divided in to a linear part and a nonlinear part, so that it can be use adaptive dual predictive control in the first part, and solve higher-order polynomials in nonlinear part. The simulations of PH titration show its better control quality.
3. Conventional control algorithms is disability and model’s changes is too slow when big jamming occurs or object change in the stable system. A new method based on the amplitude of fluctuation and force to use ADC is presented. This method use the ant-jamming of ADC, let it not only act at the beginning of the process, but improve the ability of ant-jamming in whole times.
4. ADC has some limitations now, such as shutdown phenomenon and the difficulties of parameter adjustment. We use the coordination factorβto adjust the ratio betweenpb 1 ( k )in covariance matrix and the key parameter21 ( )∧b k to avoid the shutdown. In addition, look for the relationship between the key parametersβand model, according to the different object decide the scope of parameter, in this scope we adjust the parameter easily.
1、提出了一种基于双重控制思想的广义预测自适应控制算法,该算法在模型辨识和控制的过程中,将模型的准确程度作为设计控制律的一个重要因素,在模型失配的条件下,抑制控制量的波动,同时最大限度的积累被估计参数的信息,以便最快地降低系统的不确定性。仿真结果表明,该控制算法比普通的广义预测自适应控制具有更好的控制品质,并将此方法推广到先进的预测函数控制中,效果同样令人满意。
2、提出了基于Hammerstein非线性模型的自适应双重预测控制策略。采用Hammerstein非线性模型线性化的方法,将线性部分与非线性部分有机的分离,对于线性部分作自适应双重预测控制,非线性部分当作一个求解高阶多项式的过程。通过对PH中和滴定过程的仿真实验可以看到该方法具有良好的控制品质。
3、针对系统稳态时发生大扰动或在系统变结构的情况下,自适应作用弱化,模型更新缓慢的缺点,提出了基于系统波动振幅的条件准则,强制提升双重控制作用的方案。该方案充分利用自适应双重控制所具有的良好抗干扰性能,使其抗干扰能力不仅仅在模型辨识的初始阶段发挥作用,而且在整个系统动态过程中发挥抗干扰作用。
4、针对自适应双重控制目前还存在的一些问题,比如关断现象,参数调整困难等,引入协调因子β,协调辨识得到的协方差矩阵中pb 1( k )与关键参数(b_1~2)|∧(k)的比例关系,避免由于相差悬殊而出现关断现象。另外,寻找模型中的调节参数β与模型本身的关系,确定不同被控对象下模型参数的大致范围,在一定范围内调节参数,减少工作量。
Predictive control is an advanced control algorithms based on models. It uses the forecast output and feedback correction to determine its optimal control action. Predictive control has three segments including forecasting model, rolling optimization and feedback correction. In the process of predictive computing, whether the model is matching the object is a key factor. In order to identify the parameters of the model, the adaptive control method is used. Adaptive dual control (ADC) is an advanced adaptive algorithm. It has its inherent advantages and limitations. By means of the advantages of ADC, this dissertation improves the predictive control algorithms, and reforms the limitations of the ADC, so that it can be used widely. The detail studies as follows:
1. A dual control algorithm of generalized predictive adaptive control (GPADC) is presented, it considers errors of estimated parameters sufficiently in the process of identifying model, and inhabits the fluctuation by this important factor. This algorithm accumulates the information of estimated parameters maximal at the same time. The simulation results show that this control algorithm is more effective than common generalized predictive adaptive control. Then apply adaptive dual control algorithm in predictive function control, the result is also satisfactory.
2. An adaptive dual predictive control algorithm based on nonlinear Hammerstein model is presented, using the linearization method the Hammerstein model is divided in to a linear part and a nonlinear part, so that it can be use adaptive dual predictive control in the first part, and solve higher-order polynomials in nonlinear part. The simulations of PH titration show its better control quality.
3. Conventional control algorithms is disability and model’s changes is too slow when big jamming occurs or object change in the stable system. A new method based on the amplitude of fluctuation and force to use ADC is presented. This method use the ant-jamming of ADC, let it not only act at the beginning of the process, but improve the ability of ant-jamming in whole times.
4. ADC has some limitations now, such as shutdown phenomenon and the difficulties of parameter adjustment. We use the coordination factorβto adjust the ratio betweenpb 1 ( k )in covariance matrix and the key parameter21 ( )∧b k to avoid the shutdown. In addition, look for the relationship between the key parametersβand model, according to the different object decide the scope of parameter, in this scope we adjust the parameter easily.
引文
1.孙优贤,褚键.工业过程控制技术(方法篇)[M].北京:化学工业出版社, 2006.
2.诸静.智能预测控制[M].浙江:浙江大学出版社, 2002.
3. Filatov N M, Unbehauen H. Adaptive dual control [M]. Springer, New York, 2004.
4.李清泉.自适应控制系统理论、设计与应用[M].北京:科学出版社, 1990.
5. Leonardo C. Kammer. Stability assessment for cautions iterative controller tuning [J]. automatica, 2005, 41(10): 1829-1934.
6. Campi M C, Prandini m. Randomized algorithm for the synthesis of cautious adaptive controllers [J]. Systems & Control Letters, 2003, 49(1): 21-36.
7.沈艳霞,林谨,纪志成.基于模型参考自适应方法的对偶控制方法研究[J].信息与控制, 2005, 34(4): 412-415.
8.高振斌,钱富才,刘丁.不确定随机系统LQG对偶控制[J].信息与控制, 2007, 36(1): 10-14.
9.郑言海,庄显义.具有时变不确定性系统的对偶自适应控制[J].系统工程与电子技术, 2000, 22(9): 28-30.
10. Gila E. Fruchter. A dual control problem and application to marketing [J]. European Journal of Operational Research, 2001, 130(1): 99-110.
11. Filatov N M, Unbehauen H, Keuchel U. Dual Pole-placement Controller with Direct Adaptation [J]. Automatica, 1997, 33(1): 113-117.
12. Knohl T, Xu W M, Unbehauen H. Indirect adaptive dual control for Hammerstein systems using ANN [J]. Control Engineering Practice, 2003, 11(4): 377-385.
13. Sheng J, Chen T W, Sirish L, etc. Generalized predictive control for non-uniformly sampled systems [J]. Process Control, 2002, 12: 875: 885.
14. Moon S M, Clark P L, Cole D G. The recursive generalized predictive feedback control: theory and experiments [J]. Sound and Vibration, 2005, 279: 171-199.
15.陈旦.广义预测控制及其应用研究[D]: [硕士学位论文].浙江:浙江大学工业控制技术研究所, 1996.
16. Karacan S, Hapglu H, Alpbaz M. Application of optimal adaptive generalized predictive to a packed distillation column [J]. Chemical Engineering Journal, 2001, 84(3): 389-396.
17.王国玉,韩璞,王东风.预测函数控制及其应用研究[J].系统仿真学报, 2002, 14(8): 1087-1091.
18.谢磊,张泉灵,王树.基于多模型的自适应预测函数控制[J].浙江大学学报, 2003, 37(2): 190-193.
19.石宏理,蔡远利,邱祖廉.非线性系统的递推最小二乘自适应模糊控制[J].西安交通大学学报, 2006, 40(4): 390-390.
20.袁延奇,刘文江.非线性Hammerstein系统辨识的动态分离方法[J].控制理论与应用, 2002, 19(4): 619-622.
21.黄正良,万百五,韩崇昭.辨识Hammerstein模型的两步法[J].控制理论与应用, 1995, 12(1): 34-39.
22. Zhang B, Zhang W D. Adaptive predictive function control of a class of nonlinear systems [J]. ISA Transaction, 2006, 45(2): 175-183.
23.郭键,朱瑞军,胡维礼.一类非线性系统的自适应广义预测控制[J].控制与决策, 2001, 16(3): 358-361.
24.向微,陈宗海.基于Hammerstein模型描述的非线性系统辨识新方法[J].控制理论与应用, 2007, 24(1): 143-147.
25.林卫星,张惠娣.应用离子群优化算法辨识Hammerstein模型[J].仪器仪表学报, 2006, 27(1): 75-79.
26.徐湘元,毛宗源.基于Hammerstein模型预测控制的分析与研究[J].控制理论与应用, 2000, 17(4): 529-532.
27. Bai E W, Chan K S. Identification of additive nonlinear system and its applications in generalized Hammerstein models [J]. Automatic, 2008, 44(2): 430-436.
28.张泉灵,王树青.基于Hammerstein模型的非线性预测函数控制[J].浙江大学学报,2002, 36(2): 119-122.
29. Li D, Qian F C, Fu P L. Research on Dual Control [J]. ACTA Automatic sinica, 2005, 31(1): 32-42.
30.钱富才,刘丁,李云霞.基于两极算法的对偶控制[J].控制理论与应用, 2004, 21(1): 89-93.
31. Clarke D W. adaptive predictive control [J]. Adaptive systems, 1996, 20: 83-94
32.张惠良,曲德伟,罗雄麟.基于稳态模型的自适应预测控制策略及其应用[J].化工自动化及仪表, 2006, 33(6): 11-14.
33. Froisy J B. Model predictive control-Building a bridge between theory and practice [J]. Computers & Chemical Engineering, 2006, 30(10): 1426-1435.
34.张泉灵,王树青.基于ARMAX模型自适应预测函数控制[J].信息与控制, 2000, 29(5): 431-436.
35.韩曾晋.自适应控制[M].北京:清华大学出版社, 1995.
36.张泉灵,王树青.基于神经网络的非线性预测函数控制[J].浙江大学学报, 2001,35(5): 497-501.
37.崔涛.非线性预测控制研究及其应用[D]: [硕士研究论文].浙江:浙江大学信息学院,2002.
38. Bart D S, Ton V D B. Model predictive control for max-pluslinear discrete event systems [J]. Automatica 2001, 37(1): 1049-1056.
39. Fu Y, Chai T Y. Nonlinear multivariable adaptive control using multiple models and neural networks [J]. Automatica, 2007, 43(6): 1101-1110.
40. Zhang T, Ge S S, .Hang C C. Adaptive neural network control for strict-feedback nonlinear systems using backstepping design [J]. Automatica, 2000, 36(12): 1835-1846.
41. Wang S W, Yu D L, Gomm J B. Adaptive neural network model based predictive control for air-fuel ratio of SI engines [J]. Engineering Applications of Artificial Intelligence, 2006, 19(2): 189-200.
42.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.
43. Zhang B, Li P, Zhang W D. Predictive function control of integrating process based on impulse response [J]. Control theory and Application, 2004, 2: 196-200.
44. Arbaoui M A, Vernières H L, Seguin D. Counter-current tubular heat exchanger: Modeling and adaptive predictive functional control [J]. Applied Thermal Engineering, 2007, 27(13): 2332-2338.
45.张智焕,谢磊,张泉灵.基于非线性(Wiener)模型自适应预测函数控制[J].计算机与自动化技术, 2003, 22(2): 15-17.
46. Zhao W X, Fang H T. Commentson“identification of Hammerstein nonlinear ARMAX systems”[J]. Automatic, 2007, 43(8): 1477-1478.
47. Goethals I, Pelckmans K, Johan A K. Subspace identification of Hammerstein systems using least squares support vector machines [J]. IEEE Transactions on automatic control, 2005, 50(10): 1509-1519.
48. Zhao W X. Identification for Hammerstein systems using extended least squares algorithm [C]. Proceedings of the 26th Chinese Control Conference, 2007, 26-31.
49.朱全民.非线性系统辨识[J].控制理论与应用, 1994, 11(6): 641-652.
50. Manoj S, Hasmet G, Michael N. Effect of on-line optimization techniques on model predictive control and identification (MPCI) [J]. Computers & Chemical Engineering, 2002, 26(9): 1241-1252.
51. Chyun C F, Hsun H T. Adaptive parameter identification of servo control systems with noise and high frequency uncertainties [J]. Mechanical Systems and Signal Processing,2007, 21(3): 1437-1451.
52. SukMin M, Daniel G C, Robert L.C. Real_time implementation of adaptive feedback and feedforward generalized predictive control algorithm [J]. Journal of Sound and Vibration, 2006, 294(1): 82-96.
53. Mehran M, Hassan T, Mohammad H. On tuning and complexity of an adaptive model predictive control scheduler [J]. Control Engineering Practice, 2007, 15(9): 1169-1178.
54. Danielle D, Doug C. A practical multiple model adaptive strategy for single-loop MPC [J]. Control Engineering Practice, 2003, 11(2): 141-159.
55.李辉.自适应模型算法控制的研究与仿真[D]: [硕士学位论文].河南:郑州大学电气工程学院, 2005.
56. Wang D, Li C. Self-adaptive dynamic matrix control for high-speed machining servo control [J]. The International journal of Advanced Manufacturing Technology, 2003, 21(10): 733-738.
2.诸静.智能预测控制[M].浙江:浙江大学出版社, 2002.
3. Filatov N M, Unbehauen H. Adaptive dual control [M]. Springer, New York, 2004.
4.李清泉.自适应控制系统理论、设计与应用[M].北京:科学出版社, 1990.
5. Leonardo C. Kammer. Stability assessment for cautions iterative controller tuning [J]. automatica, 2005, 41(10): 1829-1934.
6. Campi M C, Prandini m. Randomized algorithm for the synthesis of cautious adaptive controllers [J]. Systems & Control Letters, 2003, 49(1): 21-36.
7.沈艳霞,林谨,纪志成.基于模型参考自适应方法的对偶控制方法研究[J].信息与控制, 2005, 34(4): 412-415.
8.高振斌,钱富才,刘丁.不确定随机系统LQG对偶控制[J].信息与控制, 2007, 36(1): 10-14.
9.郑言海,庄显义.具有时变不确定性系统的对偶自适应控制[J].系统工程与电子技术, 2000, 22(9): 28-30.
10. Gila E. Fruchter. A dual control problem and application to marketing [J]. European Journal of Operational Research, 2001, 130(1): 99-110.
11. Filatov N M, Unbehauen H, Keuchel U. Dual Pole-placement Controller with Direct Adaptation [J]. Automatica, 1997, 33(1): 113-117.
12. Knohl T, Xu W M, Unbehauen H. Indirect adaptive dual control for Hammerstein systems using ANN [J]. Control Engineering Practice, 2003, 11(4): 377-385.
13. Sheng J, Chen T W, Sirish L, etc. Generalized predictive control for non-uniformly sampled systems [J]. Process Control, 2002, 12: 875: 885.
14. Moon S M, Clark P L, Cole D G. The recursive generalized predictive feedback control: theory and experiments [J]. Sound and Vibration, 2005, 279: 171-199.
15.陈旦.广义预测控制及其应用研究[D]: [硕士学位论文].浙江:浙江大学工业控制技术研究所, 1996.
16. Karacan S, Hapglu H, Alpbaz M. Application of optimal adaptive generalized predictive to a packed distillation column [J]. Chemical Engineering Journal, 2001, 84(3): 389-396.
17.王国玉,韩璞,王东风.预测函数控制及其应用研究[J].系统仿真学报, 2002, 14(8): 1087-1091.
18.谢磊,张泉灵,王树.基于多模型的自适应预测函数控制[J].浙江大学学报, 2003, 37(2): 190-193.
19.石宏理,蔡远利,邱祖廉.非线性系统的递推最小二乘自适应模糊控制[J].西安交通大学学报, 2006, 40(4): 390-390.
20.袁延奇,刘文江.非线性Hammerstein系统辨识的动态分离方法[J].控制理论与应用, 2002, 19(4): 619-622.
21.黄正良,万百五,韩崇昭.辨识Hammerstein模型的两步法[J].控制理论与应用, 1995, 12(1): 34-39.
22. Zhang B, Zhang W D. Adaptive predictive function control of a class of nonlinear systems [J]. ISA Transaction, 2006, 45(2): 175-183.
23.郭键,朱瑞军,胡维礼.一类非线性系统的自适应广义预测控制[J].控制与决策, 2001, 16(3): 358-361.
24.向微,陈宗海.基于Hammerstein模型描述的非线性系统辨识新方法[J].控制理论与应用, 2007, 24(1): 143-147.
25.林卫星,张惠娣.应用离子群优化算法辨识Hammerstein模型[J].仪器仪表学报, 2006, 27(1): 75-79.
26.徐湘元,毛宗源.基于Hammerstein模型预测控制的分析与研究[J].控制理论与应用, 2000, 17(4): 529-532.
27. Bai E W, Chan K S. Identification of additive nonlinear system and its applications in generalized Hammerstein models [J]. Automatic, 2008, 44(2): 430-436.
28.张泉灵,王树青.基于Hammerstein模型的非线性预测函数控制[J].浙江大学学报,2002, 36(2): 119-122.
29. Li D, Qian F C, Fu P L. Research on Dual Control [J]. ACTA Automatic sinica, 2005, 31(1): 32-42.
30.钱富才,刘丁,李云霞.基于两极算法的对偶控制[J].控制理论与应用, 2004, 21(1): 89-93.
31. Clarke D W. adaptive predictive control [J]. Adaptive systems, 1996, 20: 83-94
32.张惠良,曲德伟,罗雄麟.基于稳态模型的自适应预测控制策略及其应用[J].化工自动化及仪表, 2006, 33(6): 11-14.
33. Froisy J B. Model predictive control-Building a bridge between theory and practice [J]. Computers & Chemical Engineering, 2006, 30(10): 1426-1435.
34.张泉灵,王树青.基于ARMAX模型自适应预测函数控制[J].信息与控制, 2000, 29(5): 431-436.
35.韩曾晋.自适应控制[M].北京:清华大学出版社, 1995.
36.张泉灵,王树青.基于神经网络的非线性预测函数控制[J].浙江大学学报, 2001,35(5): 497-501.
37.崔涛.非线性预测控制研究及其应用[D]: [硕士研究论文].浙江:浙江大学信息学院,2002.
38. Bart D S, Ton V D B. Model predictive control for max-pluslinear discrete event systems [J]. Automatica 2001, 37(1): 1049-1056.
39. Fu Y, Chai T Y. Nonlinear multivariable adaptive control using multiple models and neural networks [J]. Automatica, 2007, 43(6): 1101-1110.
40. Zhang T, Ge S S, .Hang C C. Adaptive neural network control for strict-feedback nonlinear systems using backstepping design [J]. Automatica, 2000, 36(12): 1835-1846.
41. Wang S W, Yu D L, Gomm J B. Adaptive neural network model based predictive control for air-fuel ratio of SI engines [J]. Engineering Applications of Artificial Intelligence, 2006, 19(2): 189-200.
42.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.
43. Zhang B, Li P, Zhang W D. Predictive function control of integrating process based on impulse response [J]. Control theory and Application, 2004, 2: 196-200.
44. Arbaoui M A, Vernières H L, Seguin D. Counter-current tubular heat exchanger: Modeling and adaptive predictive functional control [J]. Applied Thermal Engineering, 2007, 27(13): 2332-2338.
45.张智焕,谢磊,张泉灵.基于非线性(Wiener)模型自适应预测函数控制[J].计算机与自动化技术, 2003, 22(2): 15-17.
46. Zhao W X, Fang H T. Commentson“identification of Hammerstein nonlinear ARMAX systems”[J]. Automatic, 2007, 43(8): 1477-1478.
47. Goethals I, Pelckmans K, Johan A K. Subspace identification of Hammerstein systems using least squares support vector machines [J]. IEEE Transactions on automatic control, 2005, 50(10): 1509-1519.
48. Zhao W X. Identification for Hammerstein systems using extended least squares algorithm [C]. Proceedings of the 26th Chinese Control Conference, 2007, 26-31.
49.朱全民.非线性系统辨识[J].控制理论与应用, 1994, 11(6): 641-652.
50. Manoj S, Hasmet G, Michael N. Effect of on-line optimization techniques on model predictive control and identification (MPCI) [J]. Computers & Chemical Engineering, 2002, 26(9): 1241-1252.
51. Chyun C F, Hsun H T. Adaptive parameter identification of servo control systems with noise and high frequency uncertainties [J]. Mechanical Systems and Signal Processing,2007, 21(3): 1437-1451.
52. SukMin M, Daniel G C, Robert L.C. Real_time implementation of adaptive feedback and feedforward generalized predictive control algorithm [J]. Journal of Sound and Vibration, 2006, 294(1): 82-96.
53. Mehran M, Hassan T, Mohammad H. On tuning and complexity of an adaptive model predictive control scheduler [J]. Control Engineering Practice, 2007, 15(9): 1169-1178.
54. Danielle D, Doug C. A practical multiple model adaptive strategy for single-loop MPC [J]. Control Engineering Practice, 2003, 11(2): 141-159.
55.李辉.自适应模型算法控制的研究与仿真[D]: [硕士学位论文].河南:郑州大学电气工程学院, 2005.
56. Wang D, Li C. Self-adaptive dynamic matrix control for high-speed machining servo control [J]. The International journal of Advanced Manufacturing Technology, 2003, 21(10): 733-738.