Wyniki wyszukiwania - BazTech

1

Wykorzystanie właściwości algorytmów regulacji predykcyjnej w detekcji uszkodzeń

Marusak P. M.

Pomiary Automatyka Robotyka

|

2012

|

R. 16, nr 2

338-343

PL

Algorytmy regulacji predykcyjnej dzięki sposobowi ich formułowania, w którym w umiejętny sposób wykorzystuje się model procesu, gwarantują bardzo dobrą jakość regulacji. Często ma to miejsce także w przypadku wystąpienia uszkodzenia w układzie regulacji. Wiele takich uszkodzeń może być nawet trudnych do zauważenia dzięki działaniu regulatora, jednak pożądane jest ich wykrycie. W artykule zaproponowano wykorzystanie w celu wykrywania uszkodzeń faktu, że w algorytmach regulacji predykcyjnej na bieżąco oceniana jest jakość modelu, na którym algorytm bazuje, przez porównywanie jego wyjścia z wartością wyjścia obiektu. Metoda może więc zostać użyta w algorytmach regulacji predykcyjnej dowolnego typu, bazujących zarówno na liniowych, jak i nieliniowych modelach obiektów.

EN

The model predictive control (MPC) algorithms due to their formulation and clever usage of the process model offer good control performance. It is also the case when some faults take place in the control system. Many of such faults can be even hard to notice thanks to the operation of the controller. It is, however, desired to be able to detect such situations. In the paper it is proposed to use internal signals of the MPC controllers in order to detect faults that occurred in the system. The method can be applied in the MPC algorithms based on both linear or nonlinear models.

2

Modeling of glucose concentration dynamics for predictive control of insulin administration

Radomski D., Ławryńczuk M., Marusak P. M., Tatjewski P.

Biocybernetics and Biomedical Engineering

|

2010

|

Vol. 30, no. 1

41-53

EN

The compartmental models, as Hovorka's one, are usually exact but complicated. Thus, they are not suitable for direct usage in nonlinear predictive controllers because of complexity of the resulting controller and numerical problems that may occur. Thus, simplified nonlinear (neural and fuzzy) models are developed in this paper for the future use in the predictive algorithms. Training and structure selection issues are discussed in the context of neural models. The heuristic, easy to obtain, Takagi-Sugeno fuzzy model composed of the control plant step responses is also designed. It is shown that in case of the considered biological process both nonlinear models have significantly better approximation abilities than linear ones.

3

Effective dual-mode fuzzy DMC algorithms with on-line quadratic optimization and guaranteed stability

Marusak P. M., Tatjewski P.

International Journal of Applied Mathematics and Computer Science

|

2009

|

Vol. 19, no. 1

127-141

EN

Dual-mode fuzzy dynamic matrix control (fuzzy DMC-FDMC) algorithms with guaranteed nominal stability for constrained nonlinear plants are presented. The algorithms join the advantages of fuzzy Takagi-Sugeno modeling and the predictive dual-mode approach in a computationally efficient version. Thus, they can bring an improvement in control quality compared with predictive controllers based on linear models and, at the same time, control performance similar to that obtained using more demanding algorithms with nonlinear optimization. Numerical effectiveness is obtained by using a successive linearization approach resulting in a quadratic programming problem solved on-line at each sampling instant. It is a computationally robust and fast optimization problem, which is important for on-line applications. Stability is achieved by appropriate introduction of dual-mode type stabilization mechanisms, which are simple and easy to implement. The effectiveness of the proposed approach is tested on a control system of a nonlinear plant-a distillation column with basic feedback controllers.

4

Actuator fault tolerance in control systems with predictive constrained set-point optimizers

Marusak P. M., Tatjewski P.

International Journal of Applied Mathematics and Computer Science

|

2008

|

Vol. 18, no. 4

539-551

EN

Mechanisms of fault tolerance to actuator faults in a control structure with a predictive constrained set-point optimizer are proposed. The structure considered consists of a basic feedback control layer and a local supervisory set-point optimizer which executes as frequently as the feedback controllers do with the aim to recalculate the set-points both for constraint feasibility and economic performance. The main goal of the presented reconfiguration mechanisms activated in response to an actuator blockade is to continue the operation of the control system with the fault, until it is fixed. This may be even long-term, if additional manipulated variables are available. The mechanisms are relatively simple and consist in the reconfiguration of the model structure and the introduction of appropriate constraints into the optimization problem of the optimizer, thus not affecting the numerical effectiveness. Simulation results of the presented control system for a multivariable plant are provided, illustrating the efficiency of the proposed approach.

5

Cooperation of model predictive control with steady-state economic optimisation

Ławryńczuk M., Marusak P. M., Tatjewski P.

Control and Cybernetics

|

2008

|

Vol. 37, no 1

133-158

EN

The problem of cooperation of Model Predictive Control (MPC) algorithms with steady-state economic optimisation is investigated in this paper. It is particularly important when the dynamics of disturbances is comparable with the dynamics of the process, since in such a case the classical hierarchical multilayer structure is likely to be not efficient and give the economic yield smaller than expected. This is because the economic nonlinear optimisation problem cannot be then solved on-line to update the optimal operating point as frequently as needed. On the other hand, simple target set-point optimisation based on linear models can be also insufficiently accurate. This paper introduces approximate formulations of the target set-point optimisation problem which tightly cooperates with the MPC and is solved as frequently as the MPC controller executes. Linear, linear-quadratic and piecewise-linear formulations are discussed, tuning guidelines are also given.