Project members: X.J.A. Bombois, P.M.J. Van den Hof, M. Gevers (Université Catholique de Louvain, Belgium), G. Scorletti (Université de Caen, France), B.D.O. Anderson (Australian National University and National ICT Australia, Australia), P. Date (Brunel University, United Kingdom)
Feedback control has nowadays reached many industrial sectors (chemical industry, electronic devices, high tech manufacturing industry) since this technology allows one to meet the requirements of modern industry i.e. quality, performance and optimal process operation. However, the development of feedback control is still restrained by various open issues in control and system theory. In most of these issues, the harmonious interaction between modelling (or system identification) and model-based robust control design is very important.
The control engineer often faces the problem to obtain a model and a description of the model uncertainty that are not compatible with the techniques developed for robust control design. In order to improve the design of feedback loops, the efficient connection of system identification and robust control design is thus of the highest significance and relevance. This connection must lead to an experiment-based robust control design procedure i.e. a robust control design procedure which uses the model and its related uncertainty as delivered by system identification.
Since the beginning of the nineties, many efforts have been spent in this direction. The methodology that has been developed so far can only be applied for a limited class of systems. For the practical implementation of this methodology, it is necessary to extend this class to multivariable systems with non-linear dynamic components. This extension to multivariable systems and to certain classes of non-linear systems (e.g., NARMAX systems or systems with a small non-linearity) is one objective of our research project. Besides, the very present issue of optimal experiment design also plays a very important role in our research project. The objective of optimal experiment design is to determine the less costly identification experiment that delivers sufficient information on the system dynamics for the design of a robust controller.
Next: Identification of hybrid systems Up: System identification Previous: Model based monitoring of large
Last modified: 24 March 2005, 10:16 UTC
||Search Site map|