Speaker: Alberto Bemporad, University of Siena Language: English Abstract: During the three-day period Wednesday September 22, 2003 - Friday September 24, 2003 an intensive course will be given on model predictive control and hybrid dynamical systems. The course is intended for students and engineers that want to learn about the theory and practice of Model Predictive Control (MPC) and of optimization of constrained linear systems and hybrid dynamical systems. MPC was first proposed by industry to deal with the control of multivariable systems with a large number of inputs and outputs subject to constraints. The constraints can arise from limits on manipulated variables and controlled outputs. In the last few years a theoretical basis for MPC has emerged for providing stability and robustness guarantees, for dealing with hybrid systems, and for dealing with fast sampling processes. The course will make use of the new MPC Toolbox for Matlab distributed by The MathWorks, Inc., and of the Hybrid Toolbox by A. Bemporad. There will be exercise sessions throughout the course, where students can test their understanding of the material. Outline: - General concepts of MPC. MPC based on QP. General stability properties.
- MPC Toolbox: model/problem setup, MPC objects, MPC Simulink block (exercises)
- QP/LP solvers. Multiparametric programming and Explicit MPC. MPC based on LP. Robust Explicit MPC.
- Hybrid MPC: modeling, MPC control, explicit forms, applications (exercises).
Intended audience: The course is meant for graduate students specializing in Systems Theory, Automatic Control, Process Control, Signal Processing, Optimization, Robotics, Automomous Systems, and related fields, for last year undergraduate students interested in one of the above areas, and for control engineers active in industry. Prerequisites: A basic course in automatic control and in optimization. Suitable reading: - Copies of overhead slides
- Selected scientific articles
- MPC Toolbox User's Guide
- Hybrid Toolbox User's Guide