Novel Results on Output-Feedback LQR Design
Journal article, 2023
This paper provides novel developments in output-feedback stabilization for linear time-invariant systems within the linear quadratic regulator (LQR) framework. First, we derive the necessary and sufficient conditions for output-feedback stabilizability in connection with the LQR framework. Then, we propose a novel iterative Newton's method for output-feedback LQR design and a computationally efficient modified approach that requires solving only a Lyapunov equation at each iteration step. We show that the proposed modified approach guarantees convergence from a stabilizing state-feedback to a stabilizing output-feedback solution and succeeds in solving high dimensional problems where other, state-of-the-art methods, fail. Finally, numerical examples illustrate the effectiveness of the proposed methods.
Newton method
linear time-invariant system
Regulators
linear quadratic regulator
Electrical engineering
Symmetric matrices
output-feedback
Controller design
Linear matrix inequalities
stability
Sufficient conditions
Newton's method
Convergence
Author
Adrian Ilka
Water Construction Company
Nikolce Murgovski
Chalmers, Electrical Engineering, Systems and control
IEEE Transactions on Automatic Control
0018-9286 (ISSN) 1558-2523 (eISSN)
Vol. 68 9 5187-5200Subject Categories
Computational Mathematics
Control Engineering
Signal Processing
DOI
10.1109/TAC.2022.3218560