Development of Analysis Tools for Certification of Flight Control Laws
Abstract
We have improved our ability to analyze uncertain system dynamics, including nonaffine parametric uncertainty as well as unmodeled dynamics. This entailed polytopic covering methods for graphs of vector-valued polynomial functions, and local small-gain theorems. We use simulation as a key step in aiding the nonconvex search for proofs (Lyapunov functions) and proof certificates (multipliers). Some aspects of the calculation are trivially parallelizable, and we have employed a 9-machine cluster to speed-up the analysis of uncertain systems. We also began more detailed study of systems with marginally stable linearizations (ie., adaptive systems). Finally, we made precise our claim that these techniques represent a quantitative and definitive improvement over linearized analysis.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 31, 2009
- Accession Number
- ADA531629
Entities
People
- Andrew Packard
- Gary Balas
- Pete Seiler
Organizations
- University of California, Berkeley