TK8115 - Numerisk optimalregulering
Undervises ikke studieåret 2017/2018
Advanced topics and numerical methods for optimal control problems form the core of the curriculum. This includes ways to formulate optimal control problems, numerical methods and software to solve them, and analysis of their performance and properties from a control and numerical point of view. The course is to be held next time autumn 2018.
- Thorough understanding of the methods and theoretical foundation for direct methods for formulation of nonlinear optimal control and model predictive control as numerical optimization problems, in particular related to stability, robustness, feasibility and numerical properties.
- Basic understanding of indirect methods for optimal control, such as Pontryagin´s maximum principle and the theory of dynamic programming.
- Basic understanding of numerical methods for optimization, in particular quadratic programming and nonlinear programming.
- Basic understanding of numerical linear algebra theory for solving optimization problems
- Ability to formulate well-conditioned optimization problems in standard forms
- Proficiency in use of numerical software for convex and nonlinear programming
- Knowledge of mathematical modeling tools for numerical optimal control
- Ability to select linear algebraic methods in order to exploit structural properties such as sparsity and band structure.
- Presentation of advanced scientific topics, and scientific discussion (colloquia)
- Scientific writing skills (project report)
Læringsformer og aktiviteter
The course is based on a combination of lectures, colloquia where students present topics, and a project.
TTK4130 Modelling and Simulation
TTK4135 Optimization and Control
TTK4150 Nonlinear Systems
Information will be given when the course starts.
- * Skriftlig eksamen plasseres på rom 3 dager før eksamensdato.