Journal of Software Engineering and Applications, 2012, 5, 677-681 Published Online September 2012 (
Design of the Model Predictive Control Education and
Application Interface
Eray Yilmazlar, Erkan Kaplanoğlu
Department of Mechatronics, Technical Education Faculty, Marmara University, Istanbul, Turkey.
Received June 25th, 2012; revised July 30th, 2012; accepted August 15th, 2012
In this study, an education and application interface was designed for model predictive control (MPC). For this design,
MPC Toolbox and MATLAB GUI in the MATLAB software were used. Developed interface includes model predictive
control methods, such as single-input single-output, multi-input multi-output, constrained or unconstrained systems.
The interface, developed for education of model predictive control methods, was tested in class by the students attend-
ing to the Process Dynamic and Control course.
Keywords: Model Predictive Control; MATLAB MPC; MPC Toolbox; MPC Interface
1. Introduction
In the application of automatic controllers, it is important
to realize that controller and process from a unit; credit
or discredit for results obtained are attributable to one as
much as the other. A poor controller is often able to per-
form acceptably on a process which is easy controlled.
The finest controller made, when applied to a miserably
designed process, may not deliver the desired perform-
ance. True, on badly designed process, advanced con-
trollers are able to eke out better results than older mod-
els, but on these processes, there is a definite end point
which can be approached by instrumentation and it falls
short of perfection [1].
Model predictive control (MPC) is an advanced control
method that has an important place in industrial control
[2]. While initially it was applied solely within the petro-
chemistry industry, today its use in other system control
sectors increases [3]. Considering the structure of MPC,
the controllers that generate output from the system to be
controlled on the basis of predicting and optimizing the
future behaviors of the system through mathematical mod-
els are generally called as model predictive controls [4-6].
MPC refers to a class of computer control algorithms
that utilize an explicit process model to predict the future
response of a plant. At each control interval an MPC al-
gorithm attempts to optimize future plant behavior by
computing a sequence of future manipulated variable
adjustments [7].
Today, there are many studies and innovations being
carried out in international terms concerning the applica-
tion and way of functioning of the methods of MPC [8].
As for the case in Turkey, according to the data obtained
from a TUBITAK (Scientific and Technical Research
Authority of Turkey) supported project, it was determined
that many people concerned with process control have
lack of knowledge on this matter. With the purpose of
determining the current situation of MPC education the
courses given in universities were examined and it was
found out that some MPC-related topics are included
within the scope of some postgraduate courses [9].
The using an effective experiment set or education in-
terface for showing subject will help to present of subjects
understandably and in a short time [10]. Graphical User
Interface (GUI) of a system identification device used
with MATLAB [11]. MATLAB is a well-known software
package that is widely used for control system design,
signal processing, system identification, etc. We propose
using a GUI, which is especially suitable for beginners, to
MPC [12].
On the basis of this study, it was considered that an
application and education interface providing information
on the differences, advantages and disadvantages of the
model predictive control algorithm in comparison with
other control systems and information on how and to
which systems it can be applied is needed and an interface
that runs on MATLAB environment was designed in
order to fulfill this need.
2. Place of Model Predictive Control
Education and Application Interface
within Education
Use of interfaces in education increases students’ moti-
Copyright © 2012 SciRes. JSEA
Design of the Model Predictive Control Education and Application Interface
vation through the course, helps them in developing in-
terest and positive attitude towards the course, concen-
trating their attention on the topic, transferring their
knowledge into new implementations and is useful for
increasing the permanence of information. By utilizing
these benefits of using interfaces, within the process of
lecturing MPC method, which is commonly, used in the
industry, as part of the university courses of Process Dy-
namic and Control Systems, improvement of education
quality is targeted. Also, by means of this interface the
students will acknowledge the advantages and disadvan-
tages of MPC in comparison with other control methods
through the applications they will carry out with com-
puters, and will be able to access the educational materials
providing information on the way how and the systems to
which the method can be applied. In addition, with these
applications they will be able to examine the processes of
MPC in detail and carry out analyses and evaluations [13].
3. Design and Use of Model Predictive
Control Education and Application
Operation process of the MPC method is presented in
Figure 1. The y output, which is defined as the target,
aims at the optimal operation of the system according to
the reference input. The effect of the variables affecting
this operation process, or in other words the adjustable
variables, is considerable on the output reaching the target
By means of the model predictive control method used
in controlling this effect, the variations of the adjustable
variables are predicted through the prediction horizon and
the problems these variations may cause in reaching the
desired objective is removed in the shortest time possible.
Prediction horizon indicates the process where the future
output of the system is scanned [15].
3.1. Performance of Model Predictive Control on
Matlab MPC Toolbox
MPC function runs under the MPC Toolbox of MATLAB.
Figure 1. Operation process of model predictive control.
This control method can also be conducted on MATLAB
m file or on Simulink. The study was carried out with a
user interface that runs with the codes connected to MPC
Toolbox on m file.
System block figure of the MPC process is shown in
Figure 2. Definitions of this block are given in Table 1
before starting the MPC process, the variables given in
Table 1 have to be determined [16]. The most important
ones among these are the mathematical models of the
system to be controlled and the measurable disturbances
that affect this system. After the mathematical models of
the system are developed, these models are converted into
transfer functions and this conversion is defined on
MATLAB m file [17]. Other factors that is necessary to be
defined are related with the control system such as pre-
diction horizon, control horizon, reference values and the
running duration of the process. Output equation matrix of
an interaction system is 11311 12
2321 22
as an example. Here,
11 122122
11122122 stand for the mathematical models of
the system to be controlled while
is the control signal
to be applied on this system. includes reference value,
control horizon, prediction horizon, feedback signal and
disturbance value. 1323 indicate the mathematical
model of measurable disturbances and d indicates the
numerical value of the disturbance. The data of this
equation were entered through m file and commands and
algorithms that work under MPC Toolbox were created.
These algorithms vary by each control process. For in-
stance the system being single-input single-output, multi-
input multi-output, constrained MPC, unconstrained MPC
or establishment of the modeling method as transfer
function or state space matrix changes the commands and
algorithms [18].
3.2. Interface Design on MATLAB GUI
Interfaces were designed on MATLAB GUI in order to
enable this study to be conveniently utilized in various
applications and the monitoring of the cases in different
Figure 2. Model predictive control blog as per MPC Tool-
Copyright © 2012 SciRes. JSEA
Design of the Model Predictive Control Education and Application Interface
Copyright © 2012 SciRes. JSEA
Table 1. Variables of model predictive control process.
v Immeasurable disturbance. These are the disturbances the effects on the field are not certain. MPC corrects this through feedback.
r Reference input signal.
u Adjustable variable signal. It is the signal received from the output of MPC control block.
d Measurable disturbance It is the signal included in the field and controller through feed forward in order to minimize the effect of
received measurable disturbances via predictive control.
y Output signal.
y Adjustable output signal.
z Immeasurable noise. Undesired factors such as electrical noises, sampling errors or calibration difference.
Figure 3. Interface designed for the unrestricted model predictive control of a multi-input and multi-output system.
processes for educational purpose. These designs were
again carried out through the Tools menu on GUI. It is
shown in Figure 3.
3.3. Operation of the Model Predictive Control
Education and Application Interface
When the designed interface is started, at first the entry
page is displayed as shown in Figure 4.
Five applications of the interface are presented in this
page. These are:
Unconstrained model predictive control of a single-
input single-output system.
Constrained model predictive control of a single-input
single-output system. Figure 4. Main window of the model predictive control and
application interface.
Design of the Model Predictive Control Education and Application Interface
Figure 5. Active use of the interface window designed for the unrestricted model predictive control of a multi-input and
multi-output system.
Unconstrained model predictive control of a multiple-
input multiple-output system.
Constrained model predictive control of a multiple-
input multiple-output system.
Model predictive control of a multiple-input multiple-
output system constrained with state space matrixes.
The operation of an active application is shown in
Figure 5. The sections that constitute this interface and
their tasks are as follows.
1) System Variables Button: The mathematical model
of the system and the measurable disturbance is entered
and saved into the GUI window that opens with this but-
ton as a model transfer function.
2) Control Variables Button: The reference values,
prediction and control horizon values and the values of
input and output constraints of the system to be controlled
are entered into the sub GUI window that opens by
clicking on this button.
3) Theoretical Information Button: Clicking on this
button opens the sub GUI window from where the in-
formation on the system, the MPC process and control
algorithms are displayed.
4) System and Control Information: The data saved
with the system variables and control variables buttons are
displayed separately on the interface window.
5) System Model Predictive Control Button: Clicking
on this button directs data concerning the system and
control variables to MPC Toolbox and calculation is made
with the algorithms on m file. Calculated output signal
values are shown from the graphic windows titled as ad-
justable variables and output.
4. Conclusion
In consequence of these works carried out, absorption of
the MPC methods and kinds, monitoring of the control
process and the results of the components affecting this
process and convenient testing of the designed systems
were enabled. The materials and applications to which the
model predictive control can be applied and monitored
from can be used particularly in courses concerning
process control. The study was tested together with the
students attending to the Process Dynamic and Control
course in the Mechatronics department of the the Graduate
School of Natural and Applied Sciences of Marmara
University during the semester and a concrete design was
set forth after correcting the deficiencies, from which both
the teachers and the students can monitor the process. It is
considered that the designed interface will have a sig-
nificant role in spreading this control system, which is
commonly used in the international industry, also in
[1] S. Skogestad and I. Postlethwaite, “Multivariable Feed-
back Control Analysis and Design,” John Wiley & Sons,
New York, 1996
[2] J. Richalet, “Industrial Applications of Model Based
Copyright © 2012 SciRes. JSEA
Design of the Model Predictive Control Education and Application Interface 681
Predictive Control,” Automatica, Vol. 29, No. 5, 1993, pp.
1251-1274. doi:10.1016/0005-1098(93)90049-Y
[3] J. B. Rawling, “Tutorial Overview of Model Predictive
Control,” IEEE Control Systems Magazine, Vol. 20, No.
3, 2000, pp. 38-52. doi:10.1109/37.845037
[4] E. Kaplanoglu, “Predictive Control Applications of Multi
Input Multi Output Constrained Systems,” Dissertation
Thesis, The Graduate School of Natural and Applied
Sciences of Marmara University, Istanbul, 2006
[5] D. W. Clarke, C. Mohtadi and P. S. Tuffs, “Generalized
Predictive Control—Part I. The Basic Algorithm,” Auto-
matica, Vol. 23, No. 2, 1987, pp. 137-148.
[6] H. Demircioglu and P. J. Gawthrop, “Continuous-Time
Generalized Predictive Control (CGPC),” Automatica,
Vol. 27, No. 1, 1991, pp. 55-74.
[7] S. J. Qina and T. A. Badgwellb, “A Survey of İndustrial
Model Predictive Control Technology,” Control Engine-
ering Practice, Vol. 11, No. 7, 2003, pp. 733-764.
[8] F. Temurtas, H. Temurtas, N. Yumusak and C. Oz,
“Effects of Trajectory Planning on the Model Based
Predictive Robotic Manipulator,” Control. ISCIS 2003,
LNCS 2869, 2003, pp. 545-552.
[9] E. Kaplanoglu and E. Yilmazlar, “Industrial Adaptation
for a Constrained Predictive Control Algorithm and In-
dustry Status in Turkey in Terms of Expectations,” The
Scientific and Technological Research Council of Turkey
(TÜBİTAK) 1002 Fast Support, 2011.
[10] A. Karaarslan and I. Iskender, “PFC Control Techniques
Education for Electrical and Electronics Engineering
Students Using Computer Software and Experimental
Application,” Energy Education Science and Technology
Part AEnergy Science And Research, Vol. 29, No. 1,
2012, pp. 367-380.
[11] E. Kaplanoğlu and H. Topdal, “A Simulator For Educa-
tion of Robotics,” International Journal of Electronics,
Mechanical and Mechatronics Engineering, Vol. 1, No. 1,
2009, pp. 45-48.
[12] H. Takanashi and S. Adachi, “Graphical User Interface of
System Identification Toolbox for MATLAB,” INTECH,
Croatia, 2010, p. 270.
[13] D. C. Hillman, D. J. Willis and C. N. Gunawardena,
“Learner-Interface Interaction in Distance Education: An
Extension of Contemporary Models and Strategies for
Practitioners,” The American Journal of Distance Educa-
tion, Vol. 8, No. 2, 1994, pp. 30-42.
[14] J. M. Maciejowiski, “Predictive Control with Constraints,”
Prentice Hall, Upper Saddle River, 2002.
[15] N. Ricker and M. Morari, “Model Predictive Control
Toolbox User’s Guide,” 1998.
[16] I. Yuksel, “Otomatik Kontrol,” Nobel, 2006
[17] E. F. Camacho and C. Bordons, “Model Predictive Con-
trol,” 2nd Edition, Springer-Verlag, London, 2007.
[18] D. W. Clarke and R. Scattolini, “Constrained Receding-
Horizon Predictive Control,” IEEE Proceedings, Vol. 138,
No. 4, 1991, pp. 347-354.
Copyright © 2012 SciRes. JSEA