Sequential convex programming approach for nonlinear optimization and two applications in control

Event details
Date | 06.12.2013 |
Hour | 10:15 |
Speaker |
Dr. Quoc Tran Dinh Bio: - 2001: Bsc in Applied Mathematics and Informatics, Vietnam National University, Hanoi, Vietnam. - 2003: Msc in Computer Science, Vietnam National University, Hanoi, Vietnam. - 2009-2012: PhD in Electrical Engineering, Department of Electrical Engineering and Optimization in Engineering Center (OPTEC), KU Leuven, Belgium. - 11/2012-present: Postdoc researcher at Laboratory for Information and Inference Systems, EPFL, Lausanne, Switzerland. |
Location | |
Category | Conferences - Seminars |
In this talk, we study an optimization method for solving some classes of nonlinear optimization problems, which we call “sequential convex programming” (SCP). Similar to the well-known method called “sequential quadratic programming (SQP)”, our method is an iterative scheme such that, at each iteration, it convexifies the original problem to obtain a convex subproblem and makes use of convex optimization techniques.
We first present a general algorithmic framework with local convergence guarantee (to a stationary point of the original problem). Then, we specify this framework to some special classes of nonlinear optimization problems. Finally, we show two applications of our method in control. The first application is to track the solution trajectory in nonlinear model predictive control. The second one is to design feedback laws in static output feedback control.
We first present a general algorithmic framework with local convergence guarantee (to a stationary point of the original problem). Then, we specify this framework to some special classes of nonlinear optimization problems. Finally, we show two applications of our method in control. The first application is to track the solution trajectory in nonlinear model predictive control. The second one is to design feedback laws in static output feedback control.
Practical information
- Expert
- Free