Commande robuste des systèmes en réseau
Loading...
Date
2020
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Université 20 Août 1955 - Skikda
Abstract
In recent years, the control system is becoming decentralized in location and more extensive
in scope, making it difficult to be implemented in the traditional point-to-point configuration.
Therefore, the networked control system (NCS) is introduced to replace the traditional control
due to its benefits in the wide industrial areas such as reduced installation costs, better
maintainability, and greater flexibility, where the control loop is implemented via
communication networks, and the data is transmitted between its different components
through a limit communication channel.
However, the use of communication channels in the control loop causes many problems and
constraints such as network bandwidth limit, network-induced delays, packet loss, packet
disordering, scheduling protocols, and variable sampling. These negative factors can degrade
the NCSs performance and even lead to its instability. This thesis takes into account these
problems in the analysis and the synthesis of NCSs where it analyses the robust stability and
design of a new controller for NCSs subject to network-induced delay and packet dropout.
The main contribution of this thesis is based on the reduction of the effect of network-induced
imperfections on the stability of the networked system through deriving robust less
conservative stability conditions and design a new control low ensuring the stability and good
performance of networked systems subject to the maximum possible of these imperfections.
And in addition, this work investigates the different modeling of NCSs such as stochastic,
deterministic, and hybrid that mimic reality to use it in the analysis and synthesis of
networked systems.
Finally, the stability conditions obtained are tested using numerical illustrative examples
representing different NCSs with network-induced imperfections, as well as a comparison
with other existing methods in terms of conservativeness and robustness