You are on page 1of 6

International Journal of Advanced Engineering Research and Technology (IJAERT) 284

Volume 3 Issue 8, August 2015, ISSN No.: 2348 8190

Different Configurations of Fuzzy Logic Controllers for Control of


Dynamical Systems
Mohammad Zaid1, Mohammad Ayyub2
1,2
Department of Electrical Engineering, Zakir Hussain College of Engineering & Technology,
Aligarh Muslim University Uttar Pradesh, India

Abstract
Many configurations of Fuzzy logic controllers (FLC)
are proposed in literature but some configurations are
more popular than others. This paper investigates
performance of different types Fuzzy controllers by
classifying these controllers into three broad categories
namely direct action type(DA), gain scheduled type
(GS) and hybrid type. Different plant structures
represented by their transfer functions have been taken
so that performance of each Fuzzy controller can be
judged fairly. The performance comparison is done
among these Fuzzy controllers and also with Ziegler
Nichols controller.
Keywords Fuzzy logic controller, direct action type,
gains scheduled type, hybrid type, Ziegler Nichols.

I. INTRODUCTION
Proportional- Integral-Derivative controllers are
popular and extensively used in industry because of
their robustness, simple design and they are
inexpensive. These types of controllers are best suited
for plant with first and second order processes and
where plant dynamics is already known. The problem
with conventional controllers comes when either plant
structure is unknown or if known is so complex that
design of controller by classical approach would be
impractical and cumbersome. The other problem
comes when model of a system is highly non linear or
rate of parameter change of plant is extremely high.
Fuzzy controllers perform very well in the situations
described above because by using FLC we need not to
now the plant structure and also by time needed for
design of controller may be significantly shortened.
However performance improvement using FLC will
depend on tuning and choosing a appropriate rule base
for FLC. Majority of Fuzzy controllers belong to direct
action type category, in this type [1] Fuzzy inference is
used for controlling PID actions of controller. This
paper includes four different types of Fuzzy controllers
namely conventional two input Fuzzy PI or PD
controller , two input rule coupled Fuzzy PI+PD
controller ,two input Fuzzy set point weighing
controller, and single input rule coupled Fuzzy PI

controller. In gain scheduled type FLC the Fuzzy


inference is derived or modified using additional self
tuning FLC [2] or performance based supervisory
tuning. In this paper we have included one self tuned
Fuzzy PI controller and a gain scheduled type Fuzzy
PD controller [3]. Hybrid controllers have Fuzzy logic
based proportional control [3] and may have a
conventional integral and derivative controller. This
type of controller is easily tuned and gives advantages
of both conventional and Fuzzy controllers.
Tuning of controllers whether conventional or Fuzzy is
very important. Ziegler Nichols method is generally
used for tuning of Conventional controllers. There are
no general rules for tuning of a Fuzzy logic controller
although Genetic Optimization based technique is
found to be most satisfactory for tuning of Fuzzy logic
controller. In this paper we have also designed Ziegler
Nichols controller and its performance comparison has
been done with different FLCs.

II. PRINCIPLE OF FLC


A Fuzzy logic controller design consists of four
different processes namely fuzzification, which
involves changing the crisp values at input into Fuzzy
domain. Second is design of knowledge base which
consists of design of rule base and membership
function in which expertise of control engineer may be
needed. Third is Fuzzy inference engine in which data
from external world is processed to be again used by
external world we have used Mamdani type Fuzzy
inference engine. Fourth is deffuzification in which
crisp values are obtained from Fuzzy domain. The
Fuzzy knowledge base, which reflects the collected
knowledge about how a particular control problem
must be treated, is the heart of a Fuzzy controller. The
other parts of a controller perform service tasks
necessary for the controller to be fully functional.

III. FUZZY PID STRUCTURES


A. Direct Action Fuzzy PID Controllers
1) Rule coupled Fuzzy PI controller:
This is the most commonly used Fuzzy controller.
Figure below shows the design of this controller.

www.ijaert.org

International Journal of Advanced Engineering Research and Technology (IJAERT) 285


Volume 3 Issue 8, August 2015, ISSN No.: 2348 8190

Fig.2: Surface showing relationship between e, e and


u based on rule base
Fig.1: PI type Fuzzy logic controller
th

The error signal is defined as e(k) = Set point (k


sample time) - Output (kth sample time).The change in
error is defined as e(k) = e( k) e (k -1). In the case
of a PI-type FLC, the actual value of the controller
output is obtained by u(k) = u(k-1)+ u(k), k is the
sample time and u(k) is the incremental change in
controller output. Accumulation of controller output
takes place outside the FLC and is not reflected in
rules themselves. On the other hand, if the output of
the FLC is u(k) not u(k) and there is no accumulation
of controller output, then Fig. 2 is converted into PD
type FLC.

2.) Fuzzy PI+PD controller (FPI+PD)


The Fuzzy PD+ Fuzzy PI controller (PD+PI FLC)
consists of a PD FLC in parallel with a PI FLC To
reduce the complexity of the rule-base design and gain
tuning, a common rule-base for both Fuzzy-PI and
Fuzzy-PD parts is used in this design. We have used
the same rule base for this controller as for PI
controller.

Table 1: Rule table for Fuzzy inference system

Fig.3: Structure of Fuzzy PI+PD controller


.
Table 2: Meaning of linguistic variables in Fuzzy
inference system
NVB
Negative very big
NB

Negative big

NM

Negative medium

NS
Z

Negative small
Zero

PS

Positive small

PM
PB

Positive medium
Positive big

PVB

Positive very big

It is simple to tune this type of controller compared to


three input rule coupled Fuzzy PID controller. This
type of Fuzzy PID controller has been reported to
perform better than 3 input Fuzzy controller with e
input.
3.) Fuzzy set point weighing controller(FSPWC)
This type of controller (FSPWC) was proposed by
A.Visioli consists of fuzzifying the set point. In this
way, the control law can be written as
u(k)= Kp(bysp(t)- y(t)) + Ki
+Kd

Above rule base in words can be defined as IF e is


NB and e NB then u is NVB.

here ep (t)= (bysp(t)- y(t)). In this controller instead of


fixed weight b Fuzzy controller is being used. The
rule base for this controller is same as that of Fuzzy PI
controller. In this way a simple two degree freedom
scheme is implemented; one part of the controller is
devoted to the attenuation of load disturbances, and the

www.ijaert.org

International Journal of Advanced Engineering Research and Technology (IJAERT) 286


Volume 3 Issue 8, August 2015, ISSN No.: 2348 8190

other to the set point. The use of set-point weighting


generally leads to an increase in the rise time since the
effectiveness of the proportional action is somewhat
reduced. One of the drawbacks of using this type of
controller is to tune large number of parameters.

time taken in optimization is large. Two different gain


scheduled FLCs has been included first is self tuned
Fuzzy logic controller proposed by Mudi and Pal [2]
and the second is gain scheduled PD controller.
1.) Self Tuned Fuzzy Logic controller (STFLCPI)

Fig.6: Structure of STFLCPI

Fig.4: Structure of Fuzzy set point weighing


controller
4.) Single Input Rule coupled Fuzzy PI controller
(SIFPI)
The error signal is the essential and fundamental
control component .Therefore by using the input
variable e, a one input variable e, a one input
Fuzzy PID control system is formed. This is simply
nonlinear mapping of Fuzzy proportional action
element action .This structure is the simplest of all
structures. Rule base is one dimensional as shown
below.

In this controller (STFLCPI) the effective gain of


Fuzzy controller is .Gu where is the output of Fuzzy
tuning controller and Gu is the output of main Fuzzy
controller. The output of tuned controller is not
dependent on any process parameter.

Table 3: Rule base for single input Fuzzy controller


e
NB
NS
Z
PS
PB

Fig.7: Variation of gain updating factor with error


and change of error
2.) Fuzzy Gain Scheduled PD controller
The proposed Fuzzy PD (GSFPD) [3] controller
performs adaptive tuning. Nyquist criteria is used to
determine the range [0,k] of the universe of discourse.

Fig. 5: Structure of Single input Fuzzy PI controller.


This type of controller is easy to tune as it consists of
few tuning parameters.
B.) Gain Scheduled Fuzzy Logic Controllers
Although majority of Fuzzy controllers belong to
direct type, one of the drawbacks of these types of
controllers is to tune large number of parameters and
there is no cookery book how to do it. To improve the
behavior some optimization technique can be used but
www.ijaert.org

Fig.7: Structure of GSFPD controller

International Journal of Advanced Engineering Research and Technology (IJAERT) 287


Volume 3 Issue 8, August 2015, ISSN No.: 2348 8190

C.) Hybrid Controller (HFLC)


Many schemes employing hybrid controller (HFLC)
are possible, here we have included a scheme which
employs
Fuzzy
proportional
controller
and
conventional integral controller.[4] This controller is
easily tuned and stability of the controller is not
affected.

Table 6. Performance parameters of plant two.


Controller tr(s)
O.S. ts(s) IAE
ITAE
Z.N.

0.7

60

6.5

3.1

12.17

FPI

2.97

5.8

2.586 4.225

FPI+PD

1.80

3.7

1.776 1.971

FSPWC

1.51

3.0

1.60

1.625

SIFPI

3.1

6.0

2.65

4.40

STFLC

2.44

5.0

2.161

3.27

GSFLC

2.40

4.65

2.00

3.00

HFLC

1.545

3.5

1.645 1.875

Fig.8: Structure of Hybrid controller


Table 7. Performance parameters of plant three.
Controller tr(s) O.S. ts(s) IAE
ITAE

IV. SIMULATION RESULTS


Table 4: List of different plants taken for simulation of
Fuzzy PID controllers

Z.N.

1.0

35

18

3.46

8.94

FPI

3.00

5.5

2.92

4.58

FPI+PD

2.28

4.6

1.846

2.76

FSPWC

2.0

3.6

1.664

1.94

SIFPI

4.16

7.0

3.268

4.70

STFLC

3.7

6.5

2.853

4.8

GSFLC

3.25

6.00

2.545

4.00

HFLC

1.80

5.0

1.127

1.528

Five performance parameters namely rise time (tr),


settling time (ts), peak overshoot(O.S.),absolute
integral error (IAE),absolute time integral error (ITAE)
has been taken.
Table 5. Performance parameters of plant one.
Controller tr(s)
O.S. ts(s) IAE
ITAE
Z.N.

0.55

36

6.5

1.223

.0384

FPI

1.35

3.4

1.223

1.085

FPI+PD

1.35

3.4

1.273

1.086

FSPWC

1.081

3.0

1.184

0.9753

SIFPI

1.4

3.8

0.7165

0.721

STFLC

1.72

4.0

1.40

2.00

GSFLC

0.9

2.3

0.95

1.3

HFLC

1.01

3.0

1.092

.075

Fig.9 Step response of second Plant

www.ijaert.org

International Journal of Advanced Engineering Research and Technology (IJAERT) 288


Volume 3 Issue 8, August 2015, ISSN No.: 2348 8190

Table 8. Performance parameters of plant four.


Controller tr(s) O.S. ts(s) IAE
ITAE
Z.N.

0.6

9.5

2.30

4.023

FPI

2.85

5.3

2.77

4.58

FPI+PD

1.80

3.8

1.55

1.462

FSPWC

1.6

3.7

1.131

0.98

SIFPI

2.70

5.0

2.019

2.582

STFLC

3.2

6.3

2.25

3.799

GSFLC

1.90

3.9

1.46

1.883

HFLC

1.8

3.9

1.1

1.278

Fig.11 Step response of fifth Plant

V. CONCLUSION

Fig.10 Step response of fourth Plant


Table 9. Performance parameters of plant five.
Controller tr(s) O.S. ts(s) IAE
ITAE
Z.N.

0.55

48

9.0

1.38

2.706

FPI

4.60

9.6

2.31

5.301

FPI+PD

2.26

5.0

1.109

1.277

FSPWC

.51

1.32

0.293

0.663

SIFPI

3.0

5.3

1.754

2.412

STFLC

4.4

8.0

2.143

5.176

GSFLC

1.8

3.2

1.252

1.183

HFLC

0.8

2.5

0.4652 0.221

From the above performance analysis we can


conclude that all seven configurations of Fuzzy
controller perform very well as compared to
conventional Ziegler Nichols method of tuning. Except
rise time all performance parameters show
considerable improvement over Ziegler Nichols
method of tuning. Settling time has shown
considerable reduction with the use of Fuzzy PID
controllers. The most important reduction is in
percentage peak overshoot which has been reduced to
zero in every case. Absolute integral error (IAE) and
Absolute time integral error (ITAE) has also been
reduced up to fifty percent by use of Fuzzy PID
controllers. Among the Fuzzy controller structures
Fuzzy set point controller, Fuzzy PI+PD controller,
Gain scheduled Fuzzy PD controller and Hybrid Fuzzy
controller has given best performance for every plant.
Among them the Fuzzy set point controller is found to
be more robust and is the best Fuzzy controller among
all Fuzzy controllers included in this project work.
Conventional Fuzzy PI, Self tuned FLC and Single
input FLC is found to perform better than Ziegler
Nichols Controller but they are not as good as Fuzzy
set point controller and Hybrid Fuzzy controller.
Among all Fuzzy controllers Hybrid Fuzzy controller
is most easily tuned as compared to other Fuzzy set
point controller which has many tuning parameters.
Hence it can be concluded that although Fuzzy set
point controller is the best one but the Hybrid FLC and
Gain scheduled PD FLC has also given good
performance results for each plant.

www.ijaert.org

International Journal of Advanced Engineering Research and Technology (IJAERT) 289


Volume 3 Issue 8, August 2015, ISSN No.: 2348 8190

REFERENCES
[1] George K. I. Mann, Bao-Gang Hu and G. Gosine
Analysis of Direct Action Fuzzy PID Controller
Structures , IEEE TRANSACTIONS ON SYSTEMS,
MAN,
AND
CYBERNETICSPART
B:
CYBERNETICS, VOL. 29, NO. 3, JUNE 1999.
[2] Rajani K. Mudi and Nikhil R. Pal, A Robust SelfTuning Scheme for PI- and PD-Type Fuzzy
Controllers, IEEE TRANSACTIONS ON FUZZY
SYSTEMS, VOL. 7, NO. 1, FEBRUARY 1999.
[3] Yan-Wen Huang and Pi-Cheng Tung, Design Of a
Fuzzy Gain Scheduling Controller having Input
Saturation, Journal of Marine Science and
Technology, Vol. 17, No. 4, pp. 249-256 2009.
[4] Wei Li, Design of Hybrid Fuzzy Logic
Proportional plus conventional Integral-Derivative
controller, IEEE TRANSACTIONS ON FUZZY
SYSTEMS, VOL. 6, NO. 4, NOVEMBER 1998.
[5] A.Visioli, Tuning of PID controllers with Fuzzy
logic, IEEE TRANSACTIONS ON FUZZY SYSTEMS,
OCTOBER 1995.
[6] Baogang Hu,George K. I. Mann, and Raymond G.
Gosine, New Methodology for Analytical and
Optimal Design of Fuzzy PID Controllers IEEE
TRANSACTIONS ON FUZZY SYSTEMS, VOL. 7, NO.
5, OCTOBER 1999.
[7] Baogang Hu,George K. I. Mann, and Raymond G.
Gosine, New Methodology for Analytical and
Optimal Design of Fuzzy PID Controllers IEEE
TRANSACTIONS ON FUZZY SYSTEMS, VOL. 7, NO.
5, OCTOBER 1999.

www.ijaert.org

You might also like