Welcome to the course CHEM-E7165
- (Feb 12) The second part of the assignment with FIXED BUGS is out, see `Assignments'
- (Feb 02) The second part of the assignment is out, see `Assignments'
- (Jan 17) The first part of the assignment is out, see `Assignments'
Lecturer: Francesco Corona (email@example.com - E331, Kemistintie 1)
Course content: CHEM-E7165 is an introductory course on optimal control of process systems.
We study the mathematical principles of optimal control to manipulate the dynamic behaviour of process systems and the numerics used for its solution. The course aims at bringing understanding on how to combine numerical optimisation with dynamical systems theory to formulate and solve optimal control problems in both discrete- and continuous-time. We develop the topic in general application domains in chemical and bio-chemical engineering.
- Refresher/introduction on dynamic process models and optimisation (Classes of dynamical process models; Classes of optimisation problems)
- Numerical optimisation (Optimality conditions; Newton-type algorithms) and automatic differentiation;
- Discrete-time optimal control (Formulation; Sparsity; Dynamic programming; Infinite-horizon problems; Iterative and differential);
- Continuous-time optimal control (Formulation; Hamilton-Jacobi-Bellman and Pontraygin equations; Direct and indirect methods);
- [Online optimal control (Model-predictive control; Moving horizon control and estimation)].
Learning outcomes: After the course, the participant will understand:
- Unconstrained and constrained nonlinear programming (Newton's type methods, automatic differentiation);
- Discrete-time optimal control (Dynamic programming: Basic, iterative and differential variants);
- Continuous-time optimal control (Hamilton-Jacobi-Bellman and Pontryagin approaches);
- [On-line optimal control (Model predictive control and estimation, feedback linearisation)].
Course evaluation: To pass the course the student must 1) participate to the classes and exercises and 2) pass the course project (details to be announced here, later on as the course develops).
For grading, the following conversion table applies:
|5||< -- [88,100)|
|4||< -- [76,88)|
|3||< -- [64,76)|
|2||< -- [52,64)|
|1||< -- [40,52)|
|0||< -- [00,40)|
Course material: The course is based on lecture slides/notes (to be uploaded here).
Slides/notes are mostly based on the following textbooks:
- Nocedal, J. and Wright, S. J., Nonlinear optimization, 2006;
- Bertsekas, D. P., Dynamic programing and optimal control, vol. I & II, 2017 & 2012;
- Bertsekas, D. P., Reinforcement learning and optimal control, 2019;
- Betts, J. T., Practical methods for optimal control and estimation using nonlinear programming, 2009;
- Rawlings, J. B., Mayne D. Q., Diehl, M., Model predictive control, 2017.