Computational Nonlinear Control.
Abstract
The goals of this research project were the development of control and estimation algorithms for nonlinear systems which are computationally feasible with robust performance despite numerical and modeling errors. The approach was based on the recent generalization of linear worst case (H-infinity) controllers to nonlinear systems. The construction of nonlinear H-infinity controllers depends on the solution of two PDE's of Hamilton-Jacobi type. The first is the one associated with the problem of H-infinity suboptimal control by state feedback that has appeared previously in the work of several authors. Numerical methods to compute a Taylor series solution term by term have been developed. The second PDE is a new Hamilton Jacobi equation associated with H-infinity suboptimal estimation. A hybrid computational method to solve such problems has been developed.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1997
- Accession Number
- ADA326104
Entities
People
- Arthur J. Krener
Organizations
- University of California