TY - GEN
T1 - Convex multi-objective filter optimization for output feedback L1 adaptive controller
AU - Jafarnejadsani, Hamidreza
AU - Lee, Hanmin
AU - Hovakimyan, Naira
N1 - Publisher Copyright:
© 2016, (publisher). All rights reserved.
PY - 2016
Y1 - 2016
N2 - This paper proposes a systematic analysis and synthesis method for optimal design of filter for L1 adaptive output feedback controllers. In the L1 adaptive feedback structure, the low-pass filter plays an important role in the stability and transient performance of the closed-loop system. The filter design is also the key to the trade-off between the per- formance and robustness of the closed-loop system. Although, the filter design for the L1 adaptive architecture has been studied in previous papers, the need for a numerically tractable synthesis method is yet to be fulfilled. The key challenge in establishing a satis- factory trade-off scheme between performance and robustness in L1-adaptive control archi- tecture is the non-convex nature of the underlying optimization problem for filter design. We notice that the closed-loop L1 adaptive system converges to its corresponding linear reference system both in transient and steady-state in the presence of fast estimation rates. Taking advantage of this property, we discretize the closed-loop (linear) reference system of L1 adaptive controller and formulate the filter optimization in terms of robustness to time delay and performance both in transient and steady state. Illustrative examples are provided to validate the filter design methodology of this paper.
AB - This paper proposes a systematic analysis and synthesis method for optimal design of filter for L1 adaptive output feedback controllers. In the L1 adaptive feedback structure, the low-pass filter plays an important role in the stability and transient performance of the closed-loop system. The filter design is also the key to the trade-off between the per- formance and robustness of the closed-loop system. Although, the filter design for the L1 adaptive architecture has been studied in previous papers, the need for a numerically tractable synthesis method is yet to be fulfilled. The key challenge in establishing a satis- factory trade-off scheme between performance and robustness in L1-adaptive control archi- tecture is the non-convex nature of the underlying optimization problem for filter design. We notice that the closed-loop L1 adaptive system converges to its corresponding linear reference system both in transient and steady-state in the presence of fast estimation rates. Taking advantage of this property, we discretize the closed-loop (linear) reference system of L1 adaptive controller and formulate the filter optimization in terms of robustness to time delay and performance both in transient and steady state. Illustrative examples are provided to validate the filter design methodology of this paper.
UR - http://www.scopus.com/inward/record.url?scp=85085849788&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85085849788&partnerID=8YFLogxK
U2 - 10.2514/6.2016-1375
DO - 10.2514/6.2016-1375
M3 - Conference contribution
AN - SCOPUS:85085849788
SN - 9781624103896
T3 - 2016 AIAA Guidance, Navigation, and Control Conference
BT - AIAA Guidance, Navigation, and Control Conference
T2 - AIAA Guidance, Navigation, and Control Conference, 2016
Y2 - 4 January 2016 through 8 January 2016
ER -