Authors:
Annamária R. Várkonyi-Kóczy
1
;
Attila Bencsik
1
and
Antonio Ruano
2
Affiliations:
1
Institute of Mechatronics, Óbuda University, Hungary
;
2
Faculty of Science & Technology, University of Algarve, Portugal
Keyword(s):
Power plant control, Intelligent control, Situational control, Anytime modeling, Fuzzy modeling, SVD based complexity reduction, Time critical systems, Resource insufficiency.
Related
Ontology
Subjects/Areas/Topics:
Artificial Intelligence
;
Artificial Intelligence and Decision Support Systems
;
Computational Intelligence
;
Enterprise Information Systems
;
Informatics in Control, Automation and Robotics
;
Information-Based Models for Control
;
Intelligent Control Systems and Optimization
;
Real-Time Systems Control
;
Signal Processing, Sensors, Systems Modeling and Control
;
Soft Computing
Abstract:
In time critical applications, anytime mode of operation offers a way to ensure continuous operation and to cope with the possibly dynamically changing time and resource availability. Soft Computing, especially fuzzy model based operation proved to be very advantageous in power plant control where the high complexity, nonlinearity, and possible partial knowledge usually limit the usability of classical methods. Higher Order Singular Value Decomposition based complexity reduction makes possible to convert different classes of fuzzy models into anytime models, thus offering a way to combine the advantages, like low complexity, flexibility, and robustness of fuzzy and anytime techniques. By this, a model based anytime control methodology can be suggested which is able to keep on continuous operation using non-exact, approximate models of the plant, thus preventing critical breakdowns in the operation. In this paper, an anytime modeling method is suggested which makes possible to use c
omplexity optimized fuzzy models in control. The technique is able to filter out the redundancy of fuzzy models and can determine the near optimal non-exact model of the plant considering the available time and resources. It also offers a way to improve the granularity (quality) of the model by building in new information without complexity explosion.
(More)