Conference Paper

Checking if controllers are stabilizing using closed-loop data

Res. Sch. of Inf. Sci. & Eng., Australian Nat. Univ., Canberra, ACT
DOI: 10.1109/CDC.2006.377549 In proceeding of: Decision and Control, 2006 45th IEEE Conference on
Source: IEEE Xplore

ABSTRACT Suppose an unknown plant is stabilized by a known controller. Suppose also that some knowledge of the closed-loop system is available and on the basis of that knowledge, the use of a new controller appears attractive, as may arise in iterative control and identification algorithms, and multiple-model adaptive control. The paper presents tests using a limited amount of experimental data obtained with the existing known controller for verifying that introduction of the new controller will stabilize the plant

0 Bookmarks
 · 
68 Views
  • [Show abstract] [Hide abstract]
    ABSTRACT: This paper presents a new control design approach for unknown SISO systems by using measurements. In control approaches existing in the literature, controllers are usually designed on the basis of mathematical models obtained by either using physical laws or via identification system using a set of measured data. However, due to the complex dynamics of real systems, such parametric models are only an approximation obtained after some simplifying assumptions. Therefore, the design of controllers based on a simplified model leads to a degradation in the expected performance for the closed-loop system. Our proposed approach is based on measurements to directly design controllers without going through the use of mathematical models. The principle of the proposed control methodology is to design fixed-structure controllers for which the error modulus between the closed-loop frequency response and a desired frequency response is bounded by given quantity. This problem is formulated as a nonlinear programming problem based on inequality constraints. The main advantage of our proposed approach is that the controller design is based only on a set of measurements, which allows to avoid errors associated with the identification process. Moreover, with such a proposed control method, it is guaranteed that the error between the computed and desired closed-loop frequency responses is less than a small quantity. Another feature of the proposed technique is that the structure of the controller can be selected a priori, which allows to design low-order controllers. A simulation application to measurement-based controller design for DC servomotors is presented to validate and illustrate the efficacy of the proposed approach.
    American Control Conference (ACC), 2013; 01/2013
  • [Show abstract] [Hide abstract]
    ABSTRACT: This paper deals with fixed-structure controller design for stable linear systems by using measurements. Most control design approaches developed in the literature are generally based on a mathematical model which can be obtained via identification system by using a set of measured data. However, an identified model, which is often built on the basis of some assumptions, cannot perfectly describe complex behaviors characterizing physical systems. Thus, the performance expected for the closed-loop system will be limited by the quality of such models used in the control design process. Hence, data-based controller design methods can be viewed as a possible alternative to model-based methods. In this paper, we propose to directly utilize frequency response data in the controller design. The principle is to design fixed-structure controllers for which the closed-loop frequency response fits a desired frequency response. This problem is formulated as an error minimization problem. The main feature of our proposed approach is that controller can be designed free of any mathematical model, which allows to avoid errors associated with identification process. Moreover, it enables to select low-order controllers, which are suitable for embedded systems. A simulation example is given to illustrate and validate the efficacy the proposed approach.
    Control Conference (ASCC), 2013 9th Asian; 01/2013
  • [Show abstract] [Hide abstract]
    ABSTRACT: This paper presents a new measurement-based technique to design fixed-structure controllers for unknown SISO systems. Most control design approaches existing in the literature are based on parametric models. Such models can be obtained using physical laws or via system identification using a set of measured data. Model-based control approaches have proven successful only when mathematical models used in the controller design provide a perfect description of the physical system behavior. In many practical situations, it has been shown that the derivation of models using either physical laws or system identification is usually based on simplifying assumptions due to complex dynamics characterizing real systems. Hence, the use of such simplified models in the design process may result in closed-loop performance deterioration. In this paper, an alternative approach is proposed to design controllers based on measured data without the need for model identification. Our proposed technique is novel in the sense that it can be applied even when the controller placement within a complex control system configuration is unknown. The principle of the proposed control methodology is to find the controller parameters so that the closed-loop frequency response is close to a desired frequency response. This problem is formulated as an error minimization problem. The main advantages of our proposed approach are: 1) its applicability to any complex and unknown control system configuration and without the use of any mathematical model, and 2) the design process is based on a pre-selected controller structure, which allows for the selection of low-order controllers. For simulation purposes, a PID measurement-based controller is designed to illustrate the feasibility and the efficacy of the proposed technique.
    American Control Conference (ACC), 2013; 01/2013

Full-text

View
0 Downloads
Available from