Model-based predictive control using Modelica and open source components

Abstract

This thesis is about Model Predictive Control (MPC) method for process control. It describes how this method could be implemented using some different open source software components, describing functionalities of each one and showing how the implementation has been done. Finally the code is tested to demonstrate effectiveness of this software in front of this kind of problems and to demonstrate MPC main characteristics. The main goals of this thesis are these last ones, code development and tests, so all mathematical and theoretical background are described but not as in detail as development and tests. Globally describing, MPC is a process control method where a previous knowledge of the plant is needed, so the controller have a model to simulate and predict the behavior of the system to calculate the best command signal. It has an optimization algorithm determining the optimal trajectory to bring system from initial state to desired state. Optimization is done by iterative simulation and solved online periodically at each sample time, initializing values at each time with measured feedback.