A STUDY ON THE CONVENTIONAL AND FUZZY CONTROL STEEL-CUTTING

PROCESS
S. Blent YAKUPOLU
e-mail: [email protected]

R. Nejat TUNCAY
e-mail: [email protected]

Murat YILMAZ
e-mail: [email protected]

Istanbul Technical University, Faculty of Electric-Electronics Engineering, Department of Electrical Engineering,

Electrical Machines Division,
80626, Maslak/Istanbul/Turkey
Key words: Steel-cutting, fuzzy control, simulation
ABSTRACT
This work begins with the presentation of a
conventional
steel-cutting
process,
in which
programmable logic controllers (PLC) and optical
sensors are used. Later, a new fuzzy control method is
proposed for the same application. The proposed
system is
simulated
in
MatLab/SIMULINK
environment
and
the
dynamic performance
characteristics are computed and compared with those
of experiments of the conventional system.
Encouraging results for fuzzy control application are
obtained.
I. INTRODUCTION
There have been many cutting applications in industry for
tension controlled rolling mill systems. Steel-strip cutting
is one of them. In these systems, there is a loose part to
provide buffer between two rolling mills. In Figure 1, a
schematic diagram of the system is shown. As it is seen,
there is a loose part between two drums. The purpose is to
avoid the additional tension occurring between two
drums, during cutting. Generally a dc electric motor
rotates the mill and drags the strip [1]. Another
servomotor executes the cutting process. During the
cutting process, rolling of the drag motor is interrupted
momentarily, in the mean time right hand side of the
system continues to rotate, therefore the depth of the loose
part between two drums gets higher. In opposite, when
dragging motor speed becomes high, loose part
diminishes and additional tension builds up, which may
yield damage on the strip. In order to prevent these
actions, depth of the loose part of the strip should be
detected. When the loose part of the strips depth becomes
low, motor speed goes down and loose parts bottom
moves downwards. In opposite, when the depth becomes
low, then the speed of the motor is increased to avoid
excessive loose. Without this action, loose part of the strip
may even touch the ground. Thus, dragging motors speed
should be controlled continuously.

Figure1. A PLC controlled steel cutting process.

Five optical sensors provide information about the depth
of the loose part, which is in a pit of approximately 2 m.
depth. Position information collected from the optical
sensors is transferred to a programmable logic controller
(PLC) [2]. Also motor current, voltage, power and speed
information are collected by a SCADA system. The PLC
evaluates this information and also the required position
information and then, decides on the speed of the motor
[3]. Finally, updated speed signals are produced and send
to the power-electronic circuit, which supplies the motor.
In this application Allen Bradley PLC5 programmable
logic controller has been used.
This system has some disadvantages. Firstly, optical
sensors are expensive. Secondly, since position
information is collected in five steps, position observation
is not very accurate. It has been widely accepted that,
there is a need to provide a better control with lower cost.
This paper proposes a new cutting process using fuzzy
control [4,5]. As it is known that, fuzzy control is based
upon fuzzy sets in which, sets are not defined as zero or
one. Instead, affiliation may take any value between one
and zero, which is called membership value. Control
decision is taken by fuzzy the fuzzy system, which uses
fuzzy membership functions and fuzzy rules. Finally,

whole system is defuzzyfied and required value is

obtained. There have been many successful fuzzy
applications in various process control systems. Fuzzy
control is particularly useful where human experience or
human evaluation is replaced.
In this process, whole systems are simulated in a personal
computer and fuzzy model is inserted in the simulation
model. Five optical sensors are replaced by one low cost
ultrasonic sensor. Position information of the loose part of
the strip is collected continuously and sent to the personal
computer in which fuzzy algorithm is located. Fuzzy
decision-making process is executed and required voltage
information, which is directly related with the speed of dc
motor, is produced. The detailed information of the study
is presented in the following sections.
II. DEVELOPMENT OF FUZZY CONTROL
SYSTEM
Servomotor operates in steel cutting process in start-stop
modes [6,7,8]. In other words it rotates and becomes
standstill in short intervals. Obviously, this mode of
operation adversely affects the system. Provision of a
loose part for a continuously moving strip, compensates
the effect of starts and stops. As it was explained above,
four digital sensors produce position information of the
loose part of the strip., which is a stepwise sensing. In the
proposed method, an ultrasonic sensor will be used to
monitor the position continuously, as shown in Figure 2 in

which the distance between point a and b is accurately

measured. In order to define the position error, point f is
selected as reference. Thus, error is calculated as,
e = | ag | - | af |

In order to avoid the bottom and top limits, which reflect

touching ground and complete horizontal conditions
respectively, it is assumed that error will vary between
+0.9 and 0.9 limits. Calculated (e) value will be first
input variable of the fuzzy logic controller.
It is known that, error (e) alone is not adequate; variation
speed of the error (e) should be taken into consideration.
For example, even though error is approaching zero, if
speed of it is high, then error is not going to stop at zero
error position and it is going to exceed equilibrium point
immediately. (e) is defined as the difference between
two successive error values.
e = e (t2) e (t1)

PB : variable is positive and big

(2)

e is the second input variable of the fuzzy controller.

It is known that, membership functions are important part
of fuzzy logic controller development. Particularly, rule
base system with if then rules must provide correct
information. For steel cutting process, distance and speed
of the loose part of the strip steel in the pit is chosen as
independent input variables and the following
membership functions are developed.

Figure 2. Fuzzy logic steel cutting process.

Five membership functions are defined for input and
output variables. Following definitions are necessary
for membership functions.

(1)

PO :
S :
NO :
NB :

variable is positive and medium

variable is zero
variable is negative and medium
variable is negative and big

Figure 3, 4, 5. Membership functions for distance (e), Membership functions for speed (e), Fuzzy rule base output
membership functions.
According to the operational condition at that particular
moment, fuzzy controller evaluates the situation and
produces a single output. The relationship between inputs
and a single output is established by fuzzy rules. Fuzzy
rules of a dc motor control system for steel cutting process
are given in Table 1 [9,10,11].
Table1. Fuzzy logic rules of the system.
e e

nb
no
s
po
pb

nb
nb
no
no
-

no
no
no
no
-

s
no
no
s
po
po

po
po
po
pb

system. The fuzzyfication and defuzzyfication process

and compatibility with the simulated model is inherently
achieved by using Matlab Fuzzy Logic Toolbox.
In fuzzy strip-steel cutting process, distance information
(h) showing the level of depth of the loose strip is very
important. This information is produced by an ultrasonic
sensor, which is calculated by the method described in
Figure 7.

pb
po
pb
pb

The output of the fuzzy block produces information to

reduce the error, then control system acts accordingly,
either accelerates or decelerates the DC motor. Output
membership function takes values between (+1) and
(-1). When output membership function takes (+1) value,
this means motor should accelerate immediately with a
full power. In opposite end, if output membership
function is (-1), then motor should decelerate immediately
with a full power. Any value between these limits reflects
accelerations or decelerations with a suitable power,
which is proportionally lower than maximum power
values. In order to achieve the required speed, fuzzy
output signal of which value is between (+1) and (-1)
limits, is multiplied with speed and supplied to the control
board of the servo system. Servo controller converts this
information to the voltage supply information of dc
motor. Finally, required armature voltage is produced by
power electronic supply. In Figure 6., schematic diagram
of the fuzzy controlled system is presented.
III. SIMULATION OF FUZZY STEEL CUTTING
PROCESS
An existing steel cutting process in industry, which uses a
PLC and four digital position sensors, is handled first and
fuzzy control scheme described above is replaced. The
proposed system above is modelled by using
Matlab/SIMULINK programme [12]. This model takes all
electrical and mechanical quantities into account. DC
motors armature-winding parameters, inertia and friction
coefficient are all represented in the model. The model of
ultrasonic sensor is also developed and inserted in the

Figure 6. Schematic representation of fuzzy controlled

system.
h = [ (k2/4) m2 ]

(3)

By considering the speed difference between DC motor

and servomotor, the variation of the depth of the loosestrip in the pit can be calculated as,
k(tn) = k(tn-1) + [Vdc(tn)Vsm(tn)](tntn-1)

(4)

Figure 7. Calculation of h value of the loose strip.

In order to produce real data to the simulation system,
servomotor cutting-speed values are collected by a
SCADA system and a look up table is prepared [13]. This
table is inserted in the simulation system. For dc motor
speed, initially bant speed is used as reference, later
speeds produced by fuzzy system are used and motor
speed is varied accordingly. The SIMULINK model of
fuzzy steel-cutting process is depicted in Figure 8. Threedimensional representation of the relationship between
input and output parameters are shown in Figure 9.

Figure 8. SIMULINK Model of the system.

It must be expressed that, fuzzy calculations is based upon
the simulation model. There is a possibility to get inferior
results when build and test such a system.

Figure 9. Relationship between input and output

parameters.
IV. COMPARISON BETWEEN CONVENTIONAL
AND FUZZY CONTROL
The actual industry application of a strip steel cutting
process, which uses PLC and digital position sensors, is
handled for comparison. Initially, online operational
values of the existing system have been recorded. Later,
performance characteristics of the fuzzy controlled system
have been computed and compared with those of
experiments. Actual recorded and fuzzy computed
rotational speed curves of dc motor for 33 m/s linear strip
speed are shown in Figure 10 and 11 respectively. It is
seen that, fuzzy system can respond faster to the excessive
oscillations than existing system. Same process has been
done for 61 m/s strip speed and similar results have been
obtained, which are depicted in Figure12. and 13.

Figure 10. DC motor speed variation for conventional

control system, v = 33 m/s.

Figure 11. DC motor speed variation for fuzzy control

system, v= 33 m/s.

The speed of strip steel versus time curves for fuzzy

system shows that, there will be less speed disturbances
during cutting.
Since fuzzy system provides continuous position control
and adjusts the speed accordingly it provides smooth
operation property. This would yield more reliable
operation with less waste material. Interruptions would be
less than conventional one. In addition to having better
performance characteristics, since it uses only one low
cost sensor, fuzzy controlled system will also be cost
effective.

Figure 12. DC motor speed variation for conventional

control system, v = 61 m/s.

Figure 13. DC motor speed variation for fuzzy control

system, v= 61 m/s

10. L. X. Wang, Stable Adaptive Fuzzy-Control of NonLinear Systems, IEEE Transactions on Power
Electronics, Vol. 13, No. 3, pp. 419-426, 1993.
11. M. J. Nieniewski, R. S. Marleau, Digital Simulation
of an SCR-Driven DC Motor, IEEE Transactions on
Industry Applications, Vol. 14, 1978.
12. C. M. Ong, Dynamic Simulation of Electric
Machinery Using Matlab/Simulink, Prentice Hall,
New Jersey, 1998.
13. The Math Works, Fuzzy Logic Toolbox for Use with
Matlab Users Guide, 1998.

V. CONCLUSION
A new fuzzy control method is proposed for industrial
steel-cutting process.
The simulation of this system has been achieved in
SIMULINK environment.
Performance characteristics have been computed.
Experimental performance values of existing system
have been recorded. This system uses four sensors
and a PLC to control the position of the strip-steel.
Comparison between computational and experimental
results have yielded that, proposed fuzzy system
might be superior to the existing system if it would
have been replaced. Time response is shorter, speed
control is better, possibly reliability is higher and the
cost is lower.
REFERENCES
1.
2.

3.
4.
5.
6.
7.
8.
9.

H. Edis, Doru Akm Motoruyla Hz ve Gergi

Kontrol, Yksek Lisans Tezi, stanbul Teknik
niversitesi Fen Bilimleri Enstits, stanbul, 1998.
O. Aray, Asenkron makine otomasyonu ve bir bant
tama sisteminin otomasyonunun projelendirilmesi,
Yksek Lisans Tezi, stanbul Teknik niversitesi Fen
Bilimleri Enstits, stanbul, 1999.
Allen Bredley, Digital DC Drives Troubleshooting,
1997.
J. Yen, L. A. Zadeh, R. Langari, Applications of
Fuzzy Logic and Intelligent Systems, IEEE Press,
1995.
K. Self, Designing with Fuzzy Logic, IEEE
Spectrum, Vol. 108 , pp. 42-44, 1990.
T. J. E. Miller, Brushless Permanent Magnet and
Brushless DC Motors, Clarendon Press, Oxford,
1998.
Y. Dote, S. Kinoshita, Brushless Servomotors
Fundamentals and Applications, Clarendon Press,
1990.
R. N. Tuncay, E. Tacer, G Elektronii Ders
Notlar, stanbul Teknik niversitesi, 1992.
A. Kandel, Fuzzy Mathematical Techniques with
Applications, Addison-Wesley Publishing Company,
1986.