Real –Time DC Motor Position Control by Fuzzy Logic and PID Controllers

Using Labview

Akın Delibaşı Türker Türker Galip Cansever

Yildiz Technical University Yildiz Technical University Yildiz Technical University
Dept. of Electrical Eng. Dept. of Electrical Eng. Dept. of Electrical Eng.
34349 34349 34349
adelibas@yildiz.edu.tr turker@yildiz.edu.tr cansever@yildiz.edu.tr

Abstract – This paper presents the position control of a DC rules are especially based on the knowledge of the system
motor using Fuzzy Logic and PID Control algorithms. Fuzzy behavior and the experience of the control engineer, the
Logic and PID controllers are designed based on labview FLC requires less complex mathematical modeling than
program, and the real - time position control of the DC motor classical controller does. However, to achieve high
was realized by using DAQ device. The experimental results
demonstrate that the responses of DC motor with FLC show a
performance FLCs need an effective turns scheme [1]
satisfactory, well damped control performance.

I. INTRODUCTION II. SYSTEM DESCRIPTION

There are many types of dc servo motors used in the A labview-based servo control system was built in
industries in which rotor inertia is can be very small, and order to run fuzzy and PID algorithms and also to analyze
in this result, motors with very high torque – to – inertia their works. The control system’s aims are;
ratios are commercially available [1]. Servo systems are
generally controlled by conventional Proportional –
Integral – Derivative (PID) controllers, since they designed • %0,5 or less overshot,
easily, have low cost, inexpensive maintenance and • no steady state error,
effectiveness [2]. It is necessary to know system’s • Minimum settling time,
mathematical model or to make some experiments for • Minimum rising time,
tuning PID parameters. However, it has been known that
conventional PID controllers generally do not work well
for non-linear systems, and particularly complex and vague Labview Program was used in order to develop the system
systems that have no precise mathematical models. To software. All the changes in control system can be
overcome these difficulties, various types of modified observed in real time and also user commands can be
conventional PID controllers such as auto-tuning and accepted during the process [6], [7].
adaptive PID controllers were developed lately [3], [4], [5].
Also Fuzzy Logic Controller (FLC) can be used for this
kind of problems. When compared to the conventional
controller, the main advantage of fuzzy logic is that no
mathematical modeling is required. Since the controller

Fig. 1. General block diagram of our proposed control system

A. DC Motor Model and Parameters C. Incremental Encoder and Encoder interface IC

In the control system, Pittman’s DC servo motor HP-HEDS-5310 incremental encoder has been
was used. This DC motor has the following parameters ; accoupled on dc servo motor’s rotor. However this
encoder resolution (500p/r – 0.72°) is not enough to
Torque Constant (KT) : 45.9 x 10-3Nm/A fulfill our control performance. Hence we used
Back EMF constant (KE):45.9 x 10-3V/rad/s HCTL2016 quadrature encoder interface to increase
Resistance (RT):4.62Ω resolution (2000p/r – 0.18°).
Inductance (L): 3.97mH
No – load Current (INL): 0.13A III. ANALYSIS OF PID CONTROLLER
Peak Current (IP):6.55A
Rotor inertia (J) : 5.98 x 10-4oz-in/sec2 (with encoder) Proportional – Integral – Derivative is the most
common control algorithm used in industry. We can
After neglecting frictions, system mathematical model simply express the PID Control algorithm in continuous
was obtained as follows [8] time with this equation

θ (s) KT 1
t
de
= (1) U (t ) = K C (e + ∫ edt + Td ) (2)
V (s) s ( JLs + JRs + K E KT )
2
Ti 0 dt
During the process there are some effects of noise.
B. DAQ Device Specifications Therefore we used variable filter for minimizing the
effects of noise.
National instrument PCI-1200 DAQ is a low – cost
and a multifunctional I/O device. This device allows up
to 100ks/s, 12 bit performance on 8 single-ended analog
PV f = 0.5PV + 0.25 PV (k − 1)
(3)
inputs. Besides it has the features of digital triggering + 0.175 PV (k − 2) + 0.075PV (k − 3)
capability as well as three 16-bit, 8 MHz counter/timers;
two 12-bit analog outputs; and 24 digital I/O lines.

Fig. 2. DAQ 1200 device block diagram

A. Proportional Action transfers. This algorithm prevents abrupt controller
output changes when is changed any parameter. The
U P (k ) = K C .e(k ) (4) classic tuning procedure, known as Quarter-Decay Ratio
method, was as used close loop. This system has
ultimate gain Kc, and it’s oscillation period is Tu and
B. Integral Action proportional band as PBu is 100 Kc/min
Trapezoidal integration is used to avoid sharp The PID parameters are as follows;
changes in integral action when there is a process KC = 1.67 PBu = 83.833
•
variable or setpoint pop up; the non-linear adjustment of
• Ti = 0.00031(min)
integral action causes the overshot. The larger the error
• Td = 0.0000775(min)
is the smaller the integral action becomes, as shown in
the following formula • Sample rate Ts = 2ms

IV. FUZZY LOGİC CONTROLLER

  STRUCTURE AND DESIGN (FLC)
 
K k  e(i ) + e(i − 1)   1  (5)
U I (k ) = C ∑  ∆
 t  10.e(i ) 2 The designed fuzzy logic controller has two inputs
Ti i =1  2 
1 + 2  and an output. The inputs are position error ( e ) and the
 SPrng  change of the position error ( e& ) in a sample time, and
output is the control signal ( u ).

Fig .3. The block diagram of proposed PID Controller structure

C. Derivative Action At sampling point k, the position error ( e ) and change

of the position error ( e& ) are calculated as
In the derivative action, partial derivative action was
chosen because of abrupt changes in setpoint apply only
derivative action to a filtered process variable (not the
e( k ) = θ d ( k ) − θ a ( k ) (8)
error e) to avoid derivative kick
e&(k ) = e(k ) − e(k − 1) (9)
T
U d (k ) = − K C d ( PV f ( k ) − PV f (k − 1)) (6)
∆t where θ d (k ) is the desired angular position and θ a (k )
is the actual angular position. The fuzzy controller
U (k ) = u P (k ) + U I (k ) + U D (k ) (7) consists of three stages: fuzzyfication, inference engine
and defuzzification. To define membership function e ,
The actual controller output is limited to the range e& and u the universe of discourse was divided into
specified for control output. Integral sum correction seven domains which are NB (negative big), NM
algorithm is that facilitates anti-windup and bumpless, is (negative medium), NS (negative small), Z (zero), PS
used in automatic to manual and manual to automatic (positive small), PM (positive medium) and PB (positive
big) as shown in Figure 5a, 5b, 5c.
 Fig. 4. The block diagram of proposed FLC structure

Fig. 5a. Membership functions of the position error ( e ) Fig. 5c. Membership functions of the control signal ( u )

Then, a 7x7 rule base was defined (Table 1) to develop

the inference system. Both fuzzification and inference
system were tuned experimentally. The algorithm used
for inference is max-min method, and the for
defuzzification is Center of Area (COA) method in order
to get best results.

e& / e NB NM NS Z PS PM PB
NB NB NB NB NB NS PS PB
NM NB NB NM NM Z PS PB
NS NB NB NS NS Z PM PB
Z NB NB NS Z PS PB PB
PS NB NM Z PS PS PB PB
PM NB NS Z PM PM PB PB
PB NB NS PS PB PB PB PB
Fig. 5b. Membership functions of position error’s Table 1. Rule base
change( e& )
 V. EXPERIMENTAL RESULTS

In the labview based control system that was built,

the control signal obtained from the analog output of the
DAQ device and amplified by a power amplifier in order
to feed the motor voltage. The motor was fed with ±5V
instead of ±30.8V which is nominal voltage of the
motor. To make a good comparison between Fig. 6b. Sinus responses of the controllers
performances of PID and FLC, step and half wave sinus
signals were applied to the system as control inputs. tr ts
Both the controllers’ step responses of the system from Cont. IAE ITAE %OS
(ms) (ms)
0° to 90° were given in figure 6.a. For this work, rising 57.78 1851.8
PID (38ms- (38ms- 38 4 120
times for controllers were tuned equal values in order to
Step 250ms) 250ms)
compare controllers better. As shown in figure 6.a the rsp. 34.2 772.2
position of the rotor settled at 120ms with 4% overshot FLC (38ms- (38ms- 39 0.4 80
when PID controller was used. This performance is out 250ms) 250ms)
of the control aim. If coefficients of the PID controller Sinus PID 554.83 25929
are tuned experimentally, PID controller can achieve the rsp. FLC 719.7 39919
control aim; however, after this tuning, although the
Table 2. IAE and ITAE values
overshot decreases, the settling time increases
proportionally. On the other hand desired control
purpose was achieved with 0.4% overshot and 80ms
VI. CONCLUSIONS
settling time by using FLC. Integral absolute error (IAE)
and Integral time absolute error (ITAE) values were
Most of the control systems are still based on the
calculated from 38ms to 250ms. These values are given
conventional PID controller. Because there are many
in table 2. That is seen FLC’s performance is better than
PID tuning techniques, elaborated during the last
PID controller’s performance considering IAE and ITAE
decades that make easier operator’s task. On the other
values. Also overshot and settling time for each
hand in order to tune FLC, experimental idea and
controller are given in table 2 and these values show that
experiments are needed. The development and
only the FLC achieved the control aim for the same
implementation of digital controllers for position control
rising time with PID controller’s response.
of a DC motor were successfully implemented using
labview and NI DAQ device (PCI-1200). Experimental
results show that FLC responds with less overshot and
minimum settling time.

VII. REFERENCES

[1] Khongkoom N., Kanchanathep A., Nopnakeepong S.,

Tanuthong S., Tunyasrirut S., Kagawa R., “Control of the
Fig. 6a. Step responses of the controllers position DC servo motor by fuzzy logic,” TENCON 2000.
Proceedings , Volume: 3 , 24-27 Sept. 2000 Pages:354 - 357
vol.3
Then a half wave sinus signal was applied to the system
so that system responses could be examined for [2] Tang K.S., Kim Fung Man, Guanrong Chen, Kwong S.,
trajectory following and controller’s performances could “An optimal fuzzy PID controller,” Industrial Electronics,
be compared better. Sinus responses of the system are IEEE Transactions on , Volume: 48 , Issue: 4 Aug. 2001
given in figure 6.b. System with PID type controller Pages:757 - 765
wasn’t able to follow the sinus signal and it yawed. The
reason of these deviations is integral effect of the PID [3] K. J. Aström, “Intelligent tuning,” in Adaptive Systems in
controller. From another point of view, reference signal Control and Signal Processing, L. Dugard, M. M’Saad, and I.
D. Landau, Eds. Oxford, U.K.: Pergamon, 1992, pp. 360–370.
was followed with a linearly decreasing
positive error at the first 90° and with a linearly [4] K. J. Aström, T. Hägglund, C. C. Hang, and W. K. Ho,
increasing negative error at the second 90° of the sinus “Automatic tuning and adaptation for PID controllers—A
wave by FLC. For these responses IAE and ITAE values survey,” in Adaptive Systems in Control and Signal
are given in table 2. These values show that PID Processing, L. Dugard, M. M’Saad, and I.D. Landau, Eds.
controller is better than FLC. Oxford, U.K.: Pergamon, 1992, pp. 371–376.

[5] K. J. Aström, C. C. Hang, P. Persson, andW. K. Ho,

“Toward intelligent PID control,” Automatica, vol. 28, pp. 1–
9, 1992.
[6] BridgeVIEW and Labview, “PID Control Toolkit for G
Reference Manual,” January 1998 Edition Part Number
320563B-01

[7] National Instruments, “Fuzzy Logic for G toolkit Reference

Manual,” March 1997 Edition Part Number 321511A-01

[8] Graham C. Goodwin, Stefan F. Graebe, Mario E. Salgado,

“ Control System Design, ” Prentice Hall, Upper Saddle River,
New Jersey, 2001