|Table of Contents|

Nonlinear Predictive Control for Continuous Stirred-tank Reactor Using Hammerstein Model


Research Field:
Publishing date:


Nonlinear Predictive Control for Continuous Stirred-tank Reactor Using Hammerstein Model
WANG Qing-chaoZHANG Jian-zhong
School of Energy Science and Engineering,Harbin Institute of Technology,Harbin 150001,China
nonlinear predictive control Hammerstein model least square support vector machine continuous stirred-tank reactors
A nonlinear predictive control method based on a least square support vector machine Hammerstein model is proposed for a continuous stirred-tank reactor(CSTR).A least square support vector machine(LSSVM) is used to approximate the static nonlinearity of the Hammerstein model,and a linear autoregressive model with exogenous input(ARX) is used to represent the linear block.A nonlinear predictive controller is designed based on the Hammerstein model of CSTR.The nonlinear predictive control is converted into a linear predictive control model and a nonlinear inverse model.A predictive control law is deduced,and the inverse model of the nonlinear part is constructed consequently.The proposed control scheme is compared with the traditional nonlinear predictive control and PI controller.Simulation results of CSTR indicate that the proposed nonlinear predictive controller has the highest accuracy in set point tracking of product concentration in comparison with the other two controllers.


[ 1] M orar iM, Lee J H. M ode l pred ic tive contro:l Past, present and future[ J]. Com puters and Chem ical Eng i nee ring, 1999, 23( 4): 667- 682.
[ 2] Rouhan iR, M ehra R K. M ode l algorithm control(MAC), basic theoretical properties[ J]. Autom atica, 1982, 18( 4): 401- 414.
[ 3] C larke D W, Mohtadi C, Tuffs P S. Generalized predictiv e contro.l Part 1, the basic a lgor ithm [ J]. Auto ma tica, 1987, 23( 2): 137- 148.
[ 4] 吴建国, 张培建. 基于双模型预测函数控制的PVC 聚合釜温度系统[ J]. 南京理工大学学报(自然科学版), 2005, 29( 4) : 441- 445.
[ 5] FruzzettiK P, Palazog lu A, M cDona ld K A. Nonlin ear m ode l pred ictive contro l using H amm erstein mod e ls[ J] . Journal o f Process C ontro ,l 1997, 7 ( 1): 31- 41.
[ 6] Franc isco J. Pred ictive contro l o f so lid ox ide fuel cells using fuzzyH amm erstein models[ J] . Journal of Pow er Sources, 2006, 158( 1): 245- 253.
[ 7 ] H arn ischm ache r G, M a rquardt W. Nonlinear m odel pred ictive contro l o f mu ltivar iable processes using b lockstructured m ode ls [ J ]. Contro l Eng inee ring Practice, 2007, 15( 10): 1238- 1256.
[ 8] Zhu X F, Seborg D E. Non linea r pred ic tive con tro l based on H amme rste in m ode ls [ J]. Contro l Theory and Applications, 1994, 11( 5) : 564- 575.
[ 9] Cervantes A L, Agam ennon i O E, F igueroa J L. A nonlinea r m ode l predictive control sy stem based on w iener piecew ise linearmodels[ J] . Journal of Process Contro ,l 2003, 13( 7): 655- 666.
[ 10] ArefiM M, M on tazer ia A, Poshtana J, et a.l W ienerneural identification and predictive contro l o f a m ore rea listic plug flow tubu lar reacto r[ J]. Chem ica l Eng i nee ring Journa,l 2007, 138( 1- 3): 274- 282.
[ 11] Doy le F J, Ogunna ike B A, Pearson R K. Non linear modelbased contro l using second order Vo lterra mod e ls[ J]. Autom atica, 1995, 31( 5): 697- 714.
[ 12] Venkatesw arlu C, Rao K V. Dynam ic recurrent radial basis function netwo rk model pred ictive contro l of un stab le nonlinear pro cess [ J]. Chem ical Eng inee ring Science, 2005, 60( 23): 6718- 6732.
[ 13] Bao Z J, Pi D Y, Sun Y X. Nonlinea rm ode l predictive contro l based on support vector m ach ine w ith mu ltikerne l[ J] . Ch inese Journa l o f Chem ica l Eng inee ring, 2007, 15( 5): 691- 697.
[ 14] Zhang R D, W ang S Q. Suppo rt vecto rmachine based pred ictive functiona l control design fo r output tempera ture of coking furnace[ J]. Journa l of Process Contro,l 2008, 18( 5): 439- 448.
[ 15] 刘毅, 王海清, 李平. 基于多项式核的稀疏核学习单步预测控制算法及其应用[ J]. 化工学报, 2008, 59( 10): 2541- 2545.
[ 16] Janczak A. A neura l ne tw ork approach fo r identification o fH amm erste in system [ J]. In ternational Journal o f Contro,l 2004, 76( 17) : 1749- 1766.
[ 17] Chan K H, Bao J, W h itenW J. Identification o fM I MO H amm erstein sy stem s us ing ca rdina l spline functions[ J]. Journa l o f Process Contro,l 2006, 16( 7): 659- 670.
[ 18] Goethals I, Pelckm ans K, Suykens J A K, et a.l Identifica tion ofM IMO H amm erste in m ode ls using least squa res support vec to r machines [ J ]. Autom atica, 2005, 41( 7): 1263- 1272.
[ 19] Bai EW. An optim a l tw ostag e iden tification a lgo rithm forH amm rste in W iener non linear sy stem s[ J]. Auto m atica, 1998, 34( 3): 333- 338.
[ 20] M orningred J D, Paden B E, Sebo rg D E, e t a .l An adaptiv e non linear pred ic tive controller[ J]. Chem ical Eng ineering Science, 1992, 47( 4): 755- 762.
[ 21] Chang T K, Yu D L, Yu D W. Neural network model adaption and its application to process con tro l[ J]. Ad vanced Eng ineering Inform atics, 2004, 18( 1): 1- 8.


Last Update: 2012-11-02