Much of the current body of control achieves a generic coverage of application areas by having a generic representation of the systems to be controlled. Instead of having a generic representation of systems, we propose a generic method for automatically deriving system-specific representations. We call this approach metamodelling and we use the Bond-Graph representation as a basis for this. This allows the use of particular representations for particular (possibly partially -known, possibly non-linear) systems. In this paper we illustrate the use of the bond graph representation by introducing one aspect of our approach to model-based control namely model-based observer control. The motivation for this approach comes from three areas: linear state-space observer theory, inferential control, and the application of bond graphs to observer design. An example related to process engineering is used to demonstrate the basic ideas.
[1]
Lennart Ljung,et al.
System Identification using Bond Graphs
,
1991
.
[2]
Peter J. Gawthrop,et al.
Identification of partially-known systems
,
1992,
Autom..
[3]
Dean Karnopp,et al.
Bond graphs in control: Physical state variables and observers
,
1979
.
[4]
P. J. Gawthrop,et al.
MTT: A Bond Graph Toolbox
,
1991
.
[5]
Thomas Kailath,et al.
Linear Systems
,
1980
.