Robust Identification and Control
Abstract
This final report summarizes the research contributions under AFOSR grant No. F49620-95-1-0219. The work covered two major research directions. The first is in the area of robust linear and nonlinear control. In the linear area, a complete computationally-based methodology was developed for designing controllers that can meet multiple performance objectives in both the time and frequency domain. The research culminated in a book on multi-objective control. In the nonlinear area, an alternative to gain-scheduling that requires scheduling in Lyapunov space has been proposed which gives rise to a computational tool for synthesizing controllers with guaranteed stability. In addition, the theory of Neurodynamic programming was developed to handle large-scale nonlinear optimal control problems. This research culminated in another' book on the theory and applications of Neuro-Dynamic programming. The second research direction is in the area of system identification. In that field, a new paradigm was proposed that allows deriving simple low-complexity models from noisy data obtained from complex systems. Within this paradigm, it is shown how to bridge the gap between stochastic and deterministic descriptions of noise. These developments have been shown to play a major role in many application domains.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 21, 1998
- Accession Number
- ADA356143
Entities
People
- John N. Tsitsiklis
- Munther A. Dahleh
Organizations
- Massachusetts Institute of Technology