Hierarchical Computation of Gains for the Decentralized Linear Quadratic Regulator Problem.
Abstract
The subject of this report is the application of a hierarchical computation structure to the computation of fixed decentralized state feedback gains for the regulation of linear systems with fixed (but arbitrary) dimension. This category of control problems encompasses many practical examples. In addition to those cases where the system is naturally linear, this framework extends to those nonlinear problems that can be characterized (via linearization) as linear systems driven by white noise. The decomposition technique that is utilized is that of interaction prediction. As will be shown, this particular method has the desirable characteristic of minimal computations at the supremal level. Because of the limitation of information available for feeding back under the decentralized control structure, the determination of the optimal feedback gains is dependent upon the (expected) values of the initial conditions. Thus the control which is derived is open loop. Although extensions to the basic methodology exist for forming closed loop controllers, the approach used here is totally off-line calculation of the closed loop optimal gains. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1980
- Accession Number
- ADA085763
Entities
People
- John Victor Pietras
Organizations
- University of Illinois Urbana–Champaign