This website has moved!

You will be automatically redirected to the new address. If that does not occur, visit
http://www.uscdornsifehub.com

ENH Seminar Series: Monday,March 1 "Design of optimal structured controllers for large-scaleinterconnected systems "

University of Southern California

All faculty and students are cordially invited to attend:

“Design of optimal structured controllers for large-scale interconnected systems”

by

Dr. Mihailo Jovanovic

Department of Electrical & Computer Engineering

University of Minnesota

Mihailo@umn.edu

Weekly Seminar Series on

Engineering Neuroscience & Health

Monday, March 1, 2010

4:00 p.m.

Pizza will be served from 3:30-4:00 p.m. in HNB 100

Refreshments are served starting at 3:30 in CHP 147

UPC: HNB 100 – LIVE

Hedco Neurosciences Building

UPC Campus Map/Directions:

http://www.usc.edu/about/visit/upc/

HSC: CHP 147 – Video Conference

Center for the Health Professional

HSC Campus Map/Directions:

http://www.usc.edu/about/visit/hsc/

Abstract:

We consider the design of optimal static feedback gains for interconnected systems subject to structural constraints on the distributed controller. These constraints are in the form of sparsity requirements for the feedback matrix, implying that each controller has access to information from only a limited number of subsystems. For this non-convex constrained optimal control problem, we derive necessary conditions for optimality in the form of coupled matrix equations. For stable open-loop systems, we show that in the limit of expensive control, the optimal controller can be found analytically using perturbation techniques. We use this feedback gain to initialize homotopy-based Newton iteration that finds an optimal solution to the original (non-expensive) control problem. We further employ the augmented Lagrangian method to alleviate the difficulty of finding a stabilizing structured gain to initialize efficient Newton and quasi-Newton methods that exploit the sparsity structure of the constraint set. The developed technique is used to design optimal localized controllers in large-scale vehicular formations. The issue of scaling, with respect to the number of vehicles, of global and local performance measures in the optimally-controlled formation will be discussed in detail.

Web Cast

http://capture.usc.edu/college/Catalog/pages/catalog.aspx?catalogId=946350f1-ca84-40e7-b867-e16adba01e4e

Information about all seminars can be found at

http://bme.usc.edu/valero/ENH/ENH-Schedule.html