Next: Mathematical Theory of
Up: Systems and Control
Previous: Control of Limit
Very few control design techniques can handle constraints imposed, a
priori, on the system. One of the most powerful design techniques
which can handle constraints easily and is applicable to systems in
general form is referred to as `model predictive control' or `moving
horizon control'.
Previous work in this area is concerned with construction of
stabilizing `moving horizon' control laws and design of `moving
horizon' observers.
- Moving Horizon Control Without Constraint on the Terminal
State Previous moving horizon controllers were based on the
solution of a family of open-loop optimal control problems with an
equality constraint on the terminal state (the state of the system
at the end of the optimization horizon was required to be zero). A
new class of moving horizon controllers is investigated which does
not require the satisfaction of this constraint. It is important for
applications since optimal control problems with equality
constraints are hard to solve.
- Optimization-Based Tracking of Nonlinear Systems A moving
horizon type of algorithm is investigated to achieve tracking of
reference trajectory in the output space. The algorithm is to be
tested on an aircraft model. The control objective is to generate
an on line feedback law which allows for asymptotic tracking of a
reference trajectory despite disturbances such as head-winds etc.
and model-system errors (for example, inaccurate actuator dynamics).
H. Michalska
Thierry Baron
Mon Nov 13 10:43:02 EST 1995