[Zurück]

M. Hemedi, A. Schirrer, M. Kozek:
"Comparison of LQ-optimal actuator / sensor selection approaches for flexible structure systems";
Vortrag: 9th International Conference On Mathematical Problems In Engineering, Aerospace And Sciences: ICNPAA 2012, Vienna, Austria; 10.07.2012 - 14.07.2012; in: "Vol. 1493 - 9th International Conference On Mathematical Problems In Engineering, Aerospace And Sciences: ICNPAA 2012", College Park, MD, USA (2012), ISBN: 978-0-7354-1105-0; S. 459 - 466.

In active control of flexible structures, the optimal selection of actuating andmeasurement positions on the structure
is a prerequisite to enable effective control design. This paper compares two selection onsets based on the LQG control
problem formulation. Although the two selection onsets differ considerably in their idea, it is shown at a classical bending
beam example that they produce consistent results. The methods´ characteristics are discussed and outlined, providing a
guideline for their applicability in practice. The first onset, published in [11], utilizes the LQG formalism and efficient tuning
parameter search to design tuned LQG controllers for each candidate actuator / sensor. Thereby, a normalized performance
index can be formulated which reflects the efficiency of actuation / measurement. This very general onset has been shown to
produce reliable results in numerically challenging industrial problems and is easily extended to sets of linearized systems of
nonlinear plants. The strategy presented in [3] reformulates the LQG design problem into a sensor (actuator) cost minimization
problem subject to LMI constraints which implement the corresponding Riccati equations and coupling conditions. A numeric
example for this method is given in the present paper for the first time which shows strong discrimination of non-optimal
actuator / sensor positions. Both methods are compared at a bending beam input selection task, highlighting each method´s
major advantages and limitations.

optimal input/output selection, LQG control design

http://dx.doi.org/10.1063/1.4765529