Our research is focused on how to control
intrinsically unstable structures by combining a multiagent system approach with
ideas from social science and economics. The unstable chain is a special example
of a multiagent system. It is represented by a set of mass points that are
connected by springs. This model system shows the characteristic property of
Smart Structures which is a physical coupling between neighbors within the
structure.
A very important part of our control routine is the interaction structure that
allows the system to exchange information about local displacements between
different control sections of the chain. We simulate the dynamics of the unstable
chain after initially disturbing it and compare the control performance of several
different fixed interaction structures in combination with a market mechanism to
distribute power to the actuators. We also study what kind of interaction
structure evolves if we allow the system to change the structure and to learn
how to decrease the power consumption while still controlling the unstable chain.
last changed: July 31,1997
Oliver Guenther