Description

Multiple Outputs) system is a very difficult task to perform in practice. Most often, it boils down to minimization of the coupling effects instead of full decoupling. As control of MIMO dynamic systems enjoys a continuous attention [1], [2], a challenging task becomes to include the dynamic decoupling problem [3]–[14]. In [1], an automatic two-step procedure for tuning of Proportional-Integral-Derivative (PID) controller for a Two Input Two Output (TITO) process is presented. References [6], [7] gives necessary and sufficient conditions for the existence of diagonal, block-diagonal, and triangular decoupling controllers for non-square plants and systems with non-unity feedback, and with one or two degree-offreedom controller configuration. Further, [8] has proposed a condition to check the existence of one-degree-of-freedom block decoupling controller. Parameterization of block decoupling controllers along with solving an H2 optimal problem is proposed in [9]. Reference [10] considers MIMO as proper, lumped, and linear time invariant systems and gives analytical expressions of the Input/Output (I/O) decoupling problem by the use of two-parameter stabilizing control. In [11], a robust decoupling controller for uncertain MIMO systems has been proposed, where uncertainty of model parameters and the desired performance is taken into account, and the min-max non-convex optimization problem is used in the controller design. In [15]–[18], switching, fuzzy, and neural decoupling controllers are constructed in Manuscript received 20 August, 2018; accepted 2 December, 2018. order to control the nonlinear MIMO systems. Reference [19] presents a survey on decoupling control based on multiple plant models. In recent years, technological development has increased the possibility of using predictive controllers. They seem to be ideally suited to deal with MIMO plants. Thus, we can find some works on dynamic decoupling with the use of MPC [20]–[28]. Most of them have been created for specific TITO nonlinear plants [21], [23], [25]–[27]. However, as we see in these works, the MIMO predictive controller does not automatically solve decoupling. In [21], [25]–[27] to obtain dynamic decoupling effects the MPC algorithm decelerates the change in reference signals and is changing the error weighting factors in the MPC cost function. [23] shows fuzzy, predictive, and functional control with the control law given in an analytical form. MPC with the classical control schemes are compared in this paper to show that it allows us to obtain much better results when a dynamic decoupling comes into effect. It is further illustrated that the predictive controller does not decouple plants automatically and, therefore, some tuning methods are necessary to obtain the dynamic decoupling effect. Additionally, we analyse the pros and cons of different tuning methods of the MPC algorithm to satisfy decoupling purposes. Finally, we analyse how different MPC parameters influence its performance to give some leads on how to use it to reduce loops interactions effectively. The remaining part of the paper is organized as follows. In Sections II–III, a classical and MPC approach to control MIMO plants is presented. In section IV, it is discussed how to use MPC controller to realize dynamic decoupling objectives. Pros and cons of both methods are discussed and presented in a series of simulations of a selected TITO plant in Section V. The paper ends with conclusions and some final remarks.

Instruction Files