TY - GEN
T1 - Verification of Design Specifications in L1 Adaptive Control
AU - Song, Lin
AU - Cheng, Sheng
AU - Mitra, Sayan
AU - Hovakimyan, Naira
N1 - Publisher Copyright:
© 2024 by The MITRE Corporation. Published by the American Institute of Aeronautics and Astronautics, Inc.
PY - 2024
Y1 - 2024
N2 - L1 adaptive control (L1AC) theory gives uniform steady-state and transient performance bounds on state variables and control inputs and delay margin guarantees. These guarantees are strict and can be conservative as they are derived from the small gain theorem. On the other hand, a recently-developed verification tool – Verse [1] computes over-approximations of reachable sets from simulation data for particular problem instances and can provide more realistic and less conservative performance bounds. Prior work on the verification and validation (V&V) of control design specifications and performance metrics primarily centered on optimizing design parameters through optimization or parameter variable space exploration. However, analyzing highly nonlinear systems poses challenges in mathematical analysis and performance-bound quantification, leading to a case-by-case approach to observing the design guidelines. In this paper, we propose a general framework that leverages Verse to establish verification for L1AC design specifications. Furthermore, design guidelines for closed-loop nonlinear systems with L1AC are observed from the verification results and can be instructive for control development.
AB - L1 adaptive control (L1AC) theory gives uniform steady-state and transient performance bounds on state variables and control inputs and delay margin guarantees. These guarantees are strict and can be conservative as they are derived from the small gain theorem. On the other hand, a recently-developed verification tool – Verse [1] computes over-approximations of reachable sets from simulation data for particular problem instances and can provide more realistic and less conservative performance bounds. Prior work on the verification and validation (V&V) of control design specifications and performance metrics primarily centered on optimizing design parameters through optimization or parameter variable space exploration. However, analyzing highly nonlinear systems poses challenges in mathematical analysis and performance-bound quantification, leading to a case-by-case approach to observing the design guidelines. In this paper, we propose a general framework that leverages Verse to establish verification for L1AC design specifications. Furthermore, design guidelines for closed-loop nonlinear systems with L1AC are observed from the verification results and can be instructive for control development.
UR - http://www.scopus.com/inward/record.url?scp=85193921022&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85193921022&partnerID=8YFLogxK
U2 - 10.2514/6.2024-1165
DO - 10.2514/6.2024-1165
M3 - Conference contribution
AN - SCOPUS:85193921022
SN - 9781624107115
T3 - AIAA SciTech Forum and Exposition, 2024
BT - AIAA SciTech Forum and Exposition, 2024
PB - American Institute of Aeronautics and Astronautics Inc, AIAA
T2 - AIAA SciTech Forum and Exposition, 2024
Y2 - 8 January 2024 through 12 January 2024
ER -