The control of multi-input and multi-output (MIMO) systems with multiple time delays is difficult to be done using classical control techniques, especially if the process is not open-loop stable. Model predictive control (MPC) is the natural choice in industry for this case, but its tuning and implementation are not trivial and, in general, cannot be done in the low level of the process control system. This paper aims to show a method to design MIMO filtered Smith predictors for n$\times$n processes with multiple time delays based on the decentralized direct decoupling structure. This method allows an easy tuning of the primary controller by simplifying the problem to multiple single loops if the realizability conditions presented in the paper are met. The proposed method can also be used in MPC to remove the interactions between process variables. The effectiveness of the method is illustrated through two case studies.
11 May 2016 in Journal of Control, Automation and Electrical Systems
A tuning procedure for a model predictive controller (MPC) is presented for multi-input multi-output systems. The approach consists of two steps based on a hybrid method: the goal attainment method and a variable neighborhood search. In the first step, the weights of the MPC objective function are obtained, minimizing the square error between the closed-loop response of the internal controller model and a predefined desired reference trajectory. In the second step, the integer variables of the problem (prediction and control horizons) are obtained, minimizing the square error between the closed-loop response and an optimal trajectory, aiming a controller with low computational cost and good performance. The proposed method was tested in two benchmark processes using different MPC formulations, showing satisfactory results.
This paper aims to show a methodology to design multivariable filtered Smith predictor for n × n processes with multiple time delays based on the centralized direct decoupling structure. A decoupler is introduced in this control scheme for achieving a better performance. In some cases it is necessary to introduce additional delays into the plant model transfer functions for the shortest time delay of each row of the matrix of transfer functions to appear on the main diagonal. If this condition is met, it is possible to obtain the prediction of the process output. This structure improves both the setpoint tracking and disturbance rejection responses with respect to the conventional extensions of the Smith Predictor to MIMO systems. The realizability conditions for the decoupler are provided and it is demonstrated that dead-time compensation must be enclosed in the decoupling controller matrix for simplifying the problem to multiple single-loops. The effectiveness of the method is illustrated through different simulation examples.