Advanced control strategies such as Model Predictive Control have gained wide spread interest in many areas in the chemical industries, due to fast algorithms, a well established theory and growing number of successful industrial implementations. The main feature is that the optimal control signal is determined as a constrained optimization which utilizes future predictions of the plant behaviour. Hence the controller has a plant model embedded for state estimation. The achieved closed loop performance is therefore dependent on the quality of the future predictions. The performance of the state estimator is on the other hand dependent on the accuracy of the process and the noise model. Systems with long delays in the dynamic response between the actuators and the controlled variables are notoriously difficult to control or tune. A model predictive control implementation based on a model with the correct delay will provide good set-point tracking performance as long as the prediction horizon of the controller is longer than the delay. Hence a predictive controller would perform better in rejecting known disturbances and changes between operation modes than a PI controller with time-delay compensation as e.g. a Smith predictor. A common problem for all controllers, operating on a system with a delay longer than the dominating time constant, is that a stable system may reject small disturbances before the controller have an opportunity to act. If the controller is tuned to react on these minor disturbances the change in the actuator would lead to an increase in the variance of the system output. It is therefore desired if the controller does not react on minor disturbances or measurement noise. It is on the other hand important that the controller is not detuned such that significant or sustained disturbances cannot be effectively rejected. We proposed a model predictive control implementation with a dead-band on the penalty of the tracking error as a mean to achieve good closed loop performance on time delay system. We have in simulation tested our controller on a SISO system of an industrial furnace and a MIMO system on a cement grinding circuit.
Computer-aided Chemical Engineering, 2011, p. 442-446
; Time delay systems; Model Predictive Control; Autoregressive models
Main Research Area:
Computer - Aided Chemical Engineering
21st European Symposium on Computer Aided Process Engineering, 2011