A2802 Predictive Control with Implementation

Prerequisites

Process Control, Modeling of Dynamic Systems, Linear Algebra, Programming.

Course Objectives

To achieve understanding of the importance of optimization in dynamic systems, with emphasis on Predictive Control, an understanding of how advanced control algorithmes can be implemented efficiently, how the control algorithm can be connected to a practical system/process, and the role of control algorithms in a process control system.

Learning Methods

Lectures with exercises based on Matlab with MathWorks’ Optimization Toolbox, TOMLAB, or similar tools. Group projects where a basic predictive controller is to be implemented using Matlab or a similar tool, and where the algorithm is coupled to a practical process and the system is to be controlled.

Course Description

Mathematical programming and optimization methods. Optimization of dynamic systems (overview): model fitting, state estimation, optimal control, data reconciliation. Discrete time and continuous time systems; parameterization of high order problems. Predictive control in LQ systems: DMC, GPC, state space models. Predictive control with zero control deviation; robust predictive control. Predictive control in nonlinear systems; efficient solution methods. Predictive control and controller performance. Overview of strategies for implementing control algorithms. Overview of commercial predictive controllers.

Assessment Methods

The final test and the compulsory group problem weigh 50% each in the final grade. The group project must be handed in in order to be allowed to take the final test.