ELEC97070 (EE4-54) Predictive Control
Lecturer(s): Dr Eric Kerrigan
To provide the student with an understanding of the fundamental principles in predictive control, which is the most widely used advanced control technique in industry. The emphasis of this course is on multivariable systems with constraints. This course will give an introduction to some of the formulations currently employed in industry and show how to solve challenging optimal control problems by using state-of-the-art optimisation methods. A significant component of the course includes results that have been developed in academia during the last decade, and aims to provide the student with a solid theoretical basis for understanding current and future implementations of predictive control. By the end of the course, the student should be able to design and implement their own optimisation-based controllers for the real-time control of a nonlinear system using industry-standard software.
At the end of the course, the student should be able to:
- Construct a range of ﬁnite horizon optimal control problems with constraints.
- Solve constrained ﬁnite horizon optimal control problems by formulating them as finite dimensional optimization problems.
- Explain the advantages and disadvantages of implementing the solution to ﬁnite horizon optimal control problems in a receding-, decreasing- or variable-horizon fashion.
- Translate a real-world control problem into a mathematically well-deﬁned optimal control problem.
- Design predictive controllers with guarantees of stability and feasibility.
- Write a Matlab program that implements a predictive controller
- Read relatively simple papers on predictive control and be able to extract the information from then in order to design a predictive controller.
- Make their own study notes based on the material presented during lectures and their own reading of the recommended texts and papers.
- Formulation of a variety of optimal control problems with constraints
- The receding horizon principle
- Transcribing continuous-time and sampled-data control problems into discrete-time control problems
- Solving constrained optimal control problems using numerical optimization software
- Hard and soft constraints
- Linear-quadratic predictive control
- Nonlinear predictive control
- Introduction to robust predictive control
- Stability and feasibility in receding horizon control
Exam Duration: N/A
Coursework contribution: 100%
Closed or Open Book (end of year exam): N/A
To be announced
Oral Exam Required (as final assessment): N/A
Prerequisite module(s): None required
Course Homepage: http://bb.imperial.ac.uk
Please see Module Reading list