The program is available here
Final program and book of abstracts will be available in pdf.

Plenary sessions

Fault diagnosis of dynamical processes: A soft computing approach

by Prof. Józef KORBICZ (To read the bibliography, please put the mouse cursor on the photo)
University of Zielona Góra, Institute of Control and Computation Engineering
ul. Podgórna 50, 65-246 Zielona Góra, Poland

The development of new fault diagnosis methods and techniques is forced by the increasing complexity and reliability demands of contemporary industrial systems. Taking into account many limitations of analytical methods the main objective of this paper is to present recent developments in modern fault diagnosis of dynamical systems with soft computing techniques. The concept of a model-based approach to fault detection is considered. Special emphasis is put on the problems of robust fault detection in process diagnosis with soft computing techniques such as neural networks (Group Method of Data Handing, GMDH), neuro-fuzzy networks (Takagi-Sugeno model) and evolutionary algorithms. The neural-network-based and fuzzy-based approaches are attractive for settling the fault diagnosis problem when the mathematical model is unavailable or too complex. The so-called bounded error approach is applied for defining soft computing models' uncertainty, i.e., for GMDH networks and Takagi-Sugeno fuzzy models. Finally, to illustrate the effectiveness of the GMDH and the Takagi-Sugeno networks in fault diagnosis, several powerful examples - a sugar factory value actuator (DAMADICS benchmark problem), tunnel furnace and a laboratory DC motor system fault diagnosis - are presented.

An overview of hybrid systems: analysis, control and observation

by Dr. Krishna Busawon


This talk will focus on dynamical systems in control engineering that involves both a continuous and a discrete dynamics. These are also commonly called hybrid systems. Special class of hybrid systems include impulsive and switched systems, continuous-discrete systems. Typical examples of hybrid systems include computer controlled systems, PWM-driven boost converters, the thermostat, hybrid vehicle, and much more. We shall attempt to present an overview of the current state of the art of research, in terms of control and observations. We shall discuss the problem related to stability, controllability and observability. We shall, in particular, consider the issue of observer design with discrete measurements. Several examples will be studied.

Continuous-discrete time observers for a class of MIMO nonlinear systems

by Prof. Mohamed M'Saad and Prof. Mondher Farza

GREYC UMR 6072 CNRS, ENSICAEN, Université de Caen Basse-Normandie ENSICAEN, 6 Boulevard Maréchal Juin, 14050 Caen Cedex, France

The observer design problem for nonlinear dynamical systems has received a remarkable attention over the last four decades. A considerable effort has been devoted to the observer design for systems that are observable for any input using the high gain concept for exponential convergence purposes. It is however worth mentioning that the available results are mainly devoted to the continuous-time measurements case in spite of some nice contributions concerning the sampled measurements case. The motivation of this talk consists in addressing the problem of redesigning a high gain continuous-time observer for a class of MIMO nonlinear systems that are observable for any input in the case of non uniformly sampled measurements. This provides a high gain continuous-discrete time observer witch is more suitable from an engineering practice point of view. Two design features of the proposed continuous-discrete time observer are worth to be pointed out. Firstly, the observer is particularly described by a set of differential equations with instantaneous state impulses corresponding to the measured samples and their estimates and is hence an impulsive system in nature. Secondly, the involved observer gain is time varying and its calibration is achieved throughout the tuning of two design parameters. A particular emphasis in put on the exponential convergence of the observer provided that the sampling process is not too fast in a well defined sense. Moreover, a suitable interpretation of the proposed continuous-discrete time observer is given, namely it is shown that the underlying impulsive system can be put under the form of a hybrid system witch is synthesized using a continuous-time design with an inter sample output predictor. The effectiveness of the proposed continuous-discrete time observer is emphasized throughout simulation results involving useful observer design problems. Key words : Nonlinear systems, High gain observer design, Continuous-discrete time observers, exponential convergence.





Technical Co-Sponsors

Friends of the Conference