DTU Compute Autumn 2018 Course Plan Learning Objectives A student that meet the learning objectives of the course will be able to: • Analyze and describe MPC control structures • Select processes that can be controlled by MPC • Apply convex optimization to estimation, control and system identification • Identify a linear model from data • Design and tune a linear model based estimator and predictor • Design and tune a constrained regulator • Synthesize and implement model predictive control systems • Test a linear model predictive control system by simulation • Implement Nonlinear Model Predictive Controllers • Use SQP optimization methods for NMPC • Compute solutions and sensitivities in nonlinear ODE models • Apply MPC to industrial, biomedical, and financial problems Where, When & Who • Lectures – Thursdays, 8:00 – 9:45 – Building 303A room 044 • Exercises – Thursdays, 10:00-12:00 – Building 303A room 044 • Teacher – John Bagterp Jørgensen, jbjo@dtu.dk – Office: 303B/110 – Office hours: by appointment (write an e-mail) – TAs: Dimitri Boiroux • WWW – CampusNet – www2.compute.dtu.dk/courses/02619 Evaluation and Grading • Grading will be according to the 12-scale • Grading will be based on a take-home exam project that you get at the end of the semester. • The exam project contains – An engineering case study project that you must do in groups (2-3 per group) and write a report about the project, its methods and results. Lecture Schedule 1. Introduction to computer controlled systems 2. Modeling and simulation 3. Case studies in modeling and simulation (diabetes, …) 4. Distributed models and time delays 5. System realization 6. State estimation and Kalman filtering 7. Basic MPC 8. Soft MPC and other objective functions 9. Economic MPC, stochastic MPC, financial applications 10. Nonlinear MPC 11. Nonlinear MPC
12. System identification and closed-loop simulations
13. Case studies. Handing out projects EXAM PROJECT – ENGINEERING MPC CASE STUDY