(713g) A New Extended Prediction Self-Adaptive Control (EPSAC) Strategy for Batch Control
AIChE Annual Meeting
2012
2012 AIChE Annual Meeting
Computing and Systems Technology Division
Optimization and Predictive Control II
Thursday, November 1, 2012 - 2:35pm to 2:55pm
It is well documented that linear model predictive control (MPC) techniques are predominantly used in industrial applications1. Driven by the stringent specifications on product quality, tighter environmental regulation of effluent streams, and higher competition in the process industries, the development of nonlinear model predictive control (NMPC) techniques is of interest to both the academic and industrial sectors. The main benefit of NMPC lies in its capability to handle nonlinearities and time-varying characteristics inherent in process dynamics while performing real-time dynamic optimization with constraints and bounds on both system states and manipulated variables2,3. Toward this end, various methods have been employed to approximate the process nonlinearity in the various NMPC design methods reported in the literature, including successive linearization4, neural networks5, multiple local models6,7, piecewise linearization8, andhybrid models9.
Among the aforementioned NMPC design methods, the extended prediction self-adaptive control (EPSAC) algorithm10adopted a distinct approach by approximating process variables through iterative optimization around the pre-specified base input trajectory and corresponding base output trajectory. In this manner, the outputs in the prediction horizon are obtained as the sum of a base term and a deviation term. The former is computed from the nonlinear process model based on current value of input variable given by the predefined base input trajectory, while the latter obtained from finite step response or impulse response models, from which a quadratic programming (QP) problem is formulated and can be solved iteratively11. Though successful applications of EPSAC have been reported10-13, one potential drawback of the previous EPSAC algorithms is the incorporation of finite step response or impulse models in the formulation of the control algorithm. Since model parameters are obtained by introducing a step change to the current input value specified by the base input trajectory, those model parameters obtained further away from the current time become less accurate due to process nonlinearity, leading to inevitable modeling error that degrades the achievable control performance. This shortcoming may become even worse when the EPSAC is applied in batch process control, where the objective is often to control the product quality at batch end. As a large number of step response parameters are needed at the beginning of the batch run to predict the future process outputs for the remaining batch, this eventually leads to inaccurate predicted outputs and poor control performance. Furthermore, another aim of this paper is to formulate the EPSAC algorithm to use state-space models due to its inherent flexibility to represent stable, integrating, and unstable processes1-3, 14.
Motivated by the aforementioned discussion, a new EPSAC algorithm based on the Just-in-Time Learning (JITL) method15 is developed in this work. The JITL method is a data-based modeling methodology that approximates a nonlinear system with a set of local models valid in the relevant operating regimes. In the proposed JITL-based EPSAC design, the deviation term is obtained by a set of local state-space models identified by the JITL method along the base trajectories. In simulation results of case studies for batch control of end-product quality, including an evaporative crystallizer16and a bioreactor17, the proposed EPSAC algorithm provides better closed-loop performance than its previous counterpart11.
References
[1] Qin, S.J., Badgwell, T.A. A survey of industrial model predictive control technology. Control Engineering Practice. 2003; 7(11): 733 - 764.
[2] Manenti, F. Considerations on nonlinear model predictive control techniques. Computers and Chemical Engineering. 2011; 11(35): 2491 - 2509.
[3] Darby, M.L., Nikolaou, M., MPC: Current practice and challenges. Control Engineering Practice. 2012; 4(20): 328 - 342.
[4] Lee, J.H., Ricker, N.L. Extended kalman filter based nonlinear model predictive control. Industrial & Engineering Chemistry Research. 1994; 6(33): 1530 - 1541.
[5] Peng, H., Nakano, K., Shioya, H. Nonlinear predictive control using neural nets- based local linearization ARX model-stability and industrial application. IEEE Transactions on Control Systems Technology. 2007; 1(15): 130 - 143.
[6] Dharaskar, K.P., Gupta, Y.P. Predictive control of nonlinear processes using interpolated models. Chem. Eng. Res. Design. 2000; 4(78): 573 - 580.
[7] García-Nieto, S., Martínez, M., Blasco, X., Sanchis, J. Nonlinear predictive control based on local model networks for air management in diesel engines. Control Engineering Practice. 2008; 12(16): 1399 - 1413.
[8] Cervantes, A.L., Agamennoni, O.E., Figueroa, J.L. A nonlinear model predictive control system based on Wiener piecewise linear models. Journal of Process Control. 2003; 7(13): 655 - 666.
[9] Hermanto, M.W., Braatz, R.D., Chiu, M.S. Integrated batch-to-batch and nonlinear model predictive control for polymorphic transformation in pharmaceutical crystallization. AIChE Journal. 2011; 4(57): 1008 - 1019.
[10] De Keyser, R.M.C., Cauwenberghe, A.R.V. Extended prediction self-adaptive control. In IFAC Symposium on Identification and System Parameter Estimation. 1985; 1255 - 1260.
[11] Hermanto, M.W., Chiu, M.S., Braatz, R.D. Nonlinear model predictive control for the polymorphic transformation of L-glutamic acid crystals. AIChE J. 2009; 10(55): 2631 - 2645.
[12] Gálvez-Carrillo, M., De Keyser, R., Ionescu, C., Nonlinear predictive control with dead-time compensator: Application to a solar power plant. Solar Energy. 2009; 5(83): 743 - 752.
[13] Niño, J., De Keyser, R., Syafiie, S., Ionescu, C., Struys, M., EPSAC-controlled anesthesia with online gain adaptation. International Journal of Adaptive Control and Signal Processing. 2009; (23): 455 - 471.
[14] Froisy, J.B. Model predictive control building a bridge between theory and practice. Computers and Chemical Engineering. 2006; 10-12(30): 1426 - 1435.
[15] Cheng, C., Chiu, M.S. A new data-based methodology for nonlinear process modeling. Chemical Engineering Science. 2004; 13(59): 2801 - 2810.
[16] Mesbah, A., Huesman, A.E.M., Kramer, H.J.M., Nagy, Z.K., Van de Hof, P.M.J. Real-time control of a semi-industrial fed-batch evaporative crystallizer using different direct optimization strategies. AIChE Journal. 2011; 6(57): 1557 - 1569.
[17] Peters, N., Guay, M., Dehaan, D. Real-time dynamic optimization of non-linear batch systems. Canadian Journal of Chemical Engineering. 2006; 3(84): 338 - 348.
See more of this Group/Topical: Computing and Systems Technology Division