Nonlinear Identification and Adaptive Control
Abstract
To address Air Force applications, new methods are developed for system identification (ID) and adaptive control. For linear systems, ID algorithms are developed to obtain consistent parameter estimates, stable models, and optimal inputs. Nonlinear ID methods are developed for block-structured models with measured-input nonlinearities. Subspace ID methods are used to identify linear model components, while optimization methods are used to construct efficient basis functions. Specialized methods are developed to identify nonlinear systems with output nonlinearities, limit cycle dynamics, and hysteresis. Adaptive stabilization algorithms are developed for uncertain linear and nonlinear systems under full-state feedback, as well as linear systems with unknown but bounded relative degree. Extensions to discrete-time systems are addressed. Adaptive command-following algorithms are developed for spacecraft and demonstrated on an experimental testbed. Adaptive disturbance rejection algorithms are developed for tonal and broadband disturbances. Nonlinear control algorithms are developed for shape change actuation for spacecraft. Semistability theory is developed to support research in adaptive control.
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 23, 2004
- Accession Number
- ADA421303
Entities
People
- Dennis S. Bernstein
Organizations
- University of Michigan