Marek Kubalcik

Adaptive predictive control of time-delay systems

Design of an optimal controller for higher-order or time-delay systems often leads to complex control algorithms. One of the possibilities of control of such processes is their approximation by a lower-order model with a time-delay (dead time). These time-delay processes can be effectively handled by the Model-based Predictive Control (MPC) method. The paper deals with design of an algorithm for adaptive predictive control of higher-order processes, which are approximated by a second-order model of the process with a time-delay. Most processes in industrial practice are characterized by nonlinear behavior and contain uncertainties. The adaptive control strategy is one of the possible approaches to optimal control of such systems. The proposed adaptive predictive controller for control of the time-delay system was tested and verified by simulation of a model of a laboratory heat exchanger which was obtained from measured experimental data.

Self-tuning control of nonlinear servo system: Comparison of LQ and predictive approach

The majority of processes met in the industrial practice have stochastic characteristics and eventually they embody nonlinear behaviour. Traditional controllers with fixed parameters are often unsuitable for such processes because their parameters change. The changes of process parameters are caused by changes in the manufacturing process, in the nature of the input materials, fuel, machinery use (wear) etc. Fixed controllers cannot deal with this. One possible alternative for improving the quality of control for such processes is the use of adaptive control systems. Different approaches were proposed and utilized. One successful approach is represented by self-tuning controller (STC). This approach is also called system with indirect adaptation (with direct identification). The main idea of an STC is based on the combination of a recursive identification procedure and a selected controller synthesis. In this paper, the standard STC (non-predictive) approach is verified and compared with STC based on the Model Predictive Control (MPC). The verification of both methods was implemented by the real-time control of a highly nonlinear laboratory model, the DR300 Speed Control with Variable Load.

Auto-Tuning of Digital PID Controllers Using Recursive Identification

Abstract The paper deals with algorithms for auto-tuning of digital PID controllers. Tuning is based on the identification of process model parameters using the recursive least squares method (RLSM) with directional forgetting. The parameters of the PID controllers are designed on the basis of modified Ziegler-Nichols criterion for digital control loops and pole placement approach. The process is modelled as a regression model. The relations for setting procedures are given in the form of relatively simple formulas and flow diagrams. There has been developed the MATLAB-Toolbox ATC PID for desining, testing and simulating of auto-tuning digital PID controllers.

Self-Tuning Predictive Control of Nonlinear Servo-Motor

Self-Tuning Predictive Control of Nonlinear Servo-Motor The paper is focused on a design of a self-tuning predictive model control (STMPC) algorithm and its application to a control of a laboratory servo motor. The model predictive control algorithm considers constraints of a manipulated variable. An ARX model is used in the identification part of the self-tuning controller and its parameters are recursively estimated using the recursive least squares method with the directional forgetting. The control algorithm is based on the Generalised Predictive Control (GPC) method and the optimization was realized by minimization of a quadratic and absolute values objective functions. A recursive control algorithm was designed for computation of individual predictions by incorporating a receding horizon principle. Proposed predictive controllers were verified by a real-time control of highly nonlinear laboratory model — Amira DR300.

Identification and self-tuning predictive control of heat exchanger

Heat exchange belongs to the class of basic thermal processes which occur in a range of industrial technologies, particularly in the energetic, chemical, polymer and rubber industry. The process of heat exchange is often implemented by through-flow heat exchangers. It is apparent that for an exact theoretical description of dynamics of heat exchange processes it is necessary to use partial differential equations. Heat exchange is namely a process with distributed parameters. It is also necessary to take into account its nonlinear and stochastic character. In spite of these facts, most of thermal equipment is controlled by digital modifications of PID controllers at present. This paper deals with identification of a dynamic behavior of a through-flow heat exchanger and a design of an self-tuning predictive controller for its control. The designed controller was verified by a real-time control of experimental laboratory heat exchanger.

Adaptive control of coupled-drives apparatus based on polynomial theory:

Abstract Control of a coupled-drives apparatus laboratory model as a two-input-two-output system is presented. Control algorithms based on polynomial theory and pole assignment methods are proposed. Three control algorithms are implemented and compared. Both one-degree-of-freedom and two-degree-of-freedom control system configurations were used. Problems of decoupling, where compensators suppress the interactions between control loops, were also solved. Adaptive algorithms were used to control the model. The recursive least-squares method with directional forgetting was used for the identification process. The results of real-time experiments are also included.

LQ Control Of Heat Exchanger - Design And Simulation.

Heat exchangers are devices whose primary responsibility is the transfer (exchange) of heat, typically from one fluid to another. However, they are not only used in heating applications, such as space heaters, but are also used in cooling applications, such as refrigerators and air conditioners. Heat exchange processes often contain time-delay. This paper deals with design of universal and robust digital control algorithms for control of great deal of processes with time-delay. These algorithms are realized by the digital Smith Predictor (SP) based on polynomial approach – by minimization of the Linear Quadratic (LQ) criterion. For a minimization of the LQ criterion is used spectral factorization with application of the MATLAB polynomial Toolbox. The designed polynomial digital Smith Predictors were verified in simulation conditions. Simulation model for a verification of the designed control algorithms was realized using experimental measured data on the laboratory heat exchanger. The program system MATLAB/SIMULINK was used for simulation of the designed algorithms.

Adaptive control of three — tank — system: Comparison of two methods

This paper compares two different methods applied to adaptive control of a real multivariable laboratory system of three interconnected tanks. In first case, a controller based on polynomial methods was used. The second method is based on model predictive control (MPC) approach. Both methods are based on a same model of the controlled process. Both controllers were realized as self - tuning controllers with on - line recursive identification of an ARX model of the controlled process. Results of real-time experiments are also included and quality of control achieved by both methods is compared and discussed.

Polynomial design of controllers for two-variable systems-practical implementation

Abstract This paper presents the design and simulation of adaptive control for a two input – two output system together with the real – time control of a laboratory model using this designed method. The synthesis is based on a polynomial approach. Decoupling, where the compensator is placed ahead of the system, suppresses the interactions between control loops. The results of the simulation and the real-time experiments are also given.

Predictive Control Of Two-Input Two-Output System With Non-Minimum Phase.

In this paper, a simulation of predictive control of a two-input two-output (TITO) system with nonminimum phase is presented. The proposed controller is based on extended setting of constraints. This setting represents a modification for purposes of control of nonminimum phase multivariable systems. The main problem of control of this particular type of system is undesirable undershoot in the initial phase of the control. Known methods can properly reduce the undershoot in case of predictive control of single-input single-output (SISO) systems. The paper proposes a modification of a predictive controller for two-input two-output systems with non-minimum phase behaviour. The non-minimum phase TITO models are simulated using its mathematical representation in the form of transfer function matrix.