Advanced

Real Time Model Predictive Control in JModelica.org

Ekström, Sebastian (2015)
Department of Automatic Control
Abstract
In this thesis a framework for real-time model predictive control has been developed for JModelica.org, which is an open-source platform for simulation and analysis of dynamical systems. Model predictive control (MPC) is an advanced optimizationbased control method that uses a model of the process being controlled to optimize control. The framework was tested on three different processes, real and simulated, and its performance was compared with that of an linear-quadratic regulator (LQR), which is a simpler type of controller that uses multiplication with a pre-calculated matrix to calculate the control signal from the state vector. The MPC controller was found to perform as well as or better than the LQR controller in all cases, with the... (More)
In this thesis a framework for real-time model predictive control has been developed for JModelica.org, which is an open-source platform for simulation and analysis of dynamical systems. Model predictive control (MPC) is an advanced optimizationbased control method that uses a model of the process being controlled to optimize control. The framework was tested on three different processes, real and simulated, and its performance was compared with that of an linear-quadratic regulator (LQR), which is a simpler type of controller that uses multiplication with a pre-calculated matrix to calculate the control signal from the state vector. The MPC controller was found to perform as well as or better than the LQR controller in all cases, with the main improvements being seen in the MPC controller’s ability to handle process constraints or when far from the LQR controller’s linearization point; however, the LQR controller was much faster in calculating the control signal. This also served as a first test of using JModelica.org to perform MPC on real processes, and although it performed well on the two it was tested on, further work will be needed if the MPC framework should be able to handle processes that are much faster or more
complex. (Less)
Please use this url to cite or link to this publication:
author
Ekström, Sebastian
supervisor
organization
year
type
H3 - Professional qualifications (4 Years - )
subject
ISSN
0280-5316
other publication id
ISRN LUTFD2/TFRT--5986--SE
language
English
id
7756489
date added to LUP
2015-08-10 09:16:28
date last changed
2015-08-10 09:16:28
@misc{7756489,
  abstract     = {In this thesis a framework for real-time model predictive control has been developed for JModelica.org, which is an open-source platform for simulation and analysis of dynamical systems. Model predictive control (MPC) is an advanced optimizationbased control method that uses a model of the process being controlled to optimize control. The framework was tested on three different processes, real and simulated, and its performance was compared with that of an linear-quadratic regulator (LQR), which is a simpler type of controller that uses multiplication with a pre-calculated matrix to calculate the control signal from the state vector. The MPC controller was found to perform as well as or better than the LQR controller in all cases, with the main improvements being seen in the MPC controller’s ability to handle process constraints or when far from the LQR controller’s linearization point; however, the LQR controller was much faster in calculating the control signal. This also served as a first test of using JModelica.org to perform MPC on real processes, and although it performed well on the two it was tested on, further work will be needed if the MPC framework should be able to handle processes that are much faster or more
complex.},
  author       = {Ekström, Sebastian},
  issn         = {0280-5316},
  language     = {eng},
  note         = {Student Paper},
  title        = {Real Time Model Predictive Control in JModelica.org},
  year         = {2015},
}