See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/260122357

Real-Time Performance Evaluation of a Fuzzy PI + Fuzzy PD Controller for

Liquid-Level Process

Article in International Journal of Intelligent Control and Systems · June 2008

CITATIONS
READS
51
1,296

3 authors, including:

Vineet Kumar
K. P. S. Rana
Netaji Subhas University of Technology
Netaji Subhas University of Technology
142 PUBLICATIONS 2,146 CITATIONS
131 PUBLICATIONS 2,026 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Modeling, Diagnosis & Control of Non-Linear Process Using Soft-Computing Techniques View project

Sensors View project

All content following this page was uploaded by Vineet Kumar on 12 February 2014.
The user has requested enhancement of the downloaded file.
INTERNATIONAL JOURNAL OF INTELLIGENT CONTROL AND SYSTEMS
VOL. 13, NO. 2, JUNE 2008, 89-96

Real-Time Performance Evaluation of a Fuzzy PI + Fuzzy PD

Controller for Liquid-Level Process

Vineet KUMAR, K.P.S. RANA and Vandna GUPTA

Abstract- In this paper, a comparative study was carried out to

evaluate the real-time performance of fuzzy proportional-integral loops: an inner loop with fast dynamic to eliminate input
plus fuzzy proportional-derivative (Fuzzy PI + Fuzzy PD) controller disturbances, while the outer loop to regulate output
with the real-time performance of Conventional PI for a liquid-level performance [2,7].
process experiment. The process considered for this experiment
shows highly nonlinear behavior due to equal percentage pneumatic
However, conventional PID controllers cannot
control valve. NATIONAL INSTRUMENTS™ based hardware and provide a general solution to all control problems. The
software tools (LabVIEW™) were used for precise and accurate processes involved are in general complex and time-
acquisition, measurement and control. The real time variant, with delays and non-linearity, and often with
implementation of the Fuzzy PI + Fuzzy PD controller was carried
out in two configurations: namely, feedback and cascade. In cascade
poorly defined dynamics. When the process becomes too
control configuration Fuzzy PI + Fuzzy PD controller was complex to be described by analytical models, it is
implemented in the primary loop. The secondary loop was tuned unlikely to be efficiently controlled by conventional
using the conventional PI controller. It was evaluated that Fuzzy approaches. To overcome these difficulties, various
controller perform better in comparison with conventional
controller in both the feedback and cascade control configurations.
types of modified conventional PID controllers such as
autotuning and adaptive PID controllers were developed
Index Terms— Fuzzy PI + Fuzzy PD, PID, Cascade Control. lately [6,17]. Also, a class of nonconventional type of
Process Control. PID controller employing fuzzy logic has been designed
and simulated for this purpose [6,9,11].
1. INTRODUCTION Fuzzy Logic Controller (FLC) has emerged as one of
the most active and useful research areas in the fuzzy
Intelligent Process Control methods such as fuzzy logic control theory. That is why fuzzy logic controllers have
control have shown some success, there is a significant been successfully applied for control of various physical
need to evaluate their real time performance relative to processes. Basically there are two approaches to a fuzzy
conventional control approaches, particularly in an controller design: an expert approach and a control
experimental setting. Such evaluations help to determine engineering approach. In the first, the fuzzy controller
the performance of the new intelligent process control structure and parameters choice are assumed to be the
methods, and provide the engineer with general responsibility of the experts. Consequently, design and
guidelines on how to apply them to more complex real- performance of a fuzzy controller depend mainly on the
world applications [12-15]. Despite a lot of research and knowledge and experience of the experts, or intuition
the large number of different solutions proposed, most and professional feeling of a designer. This dependence,
industrial control systems are still based on conventional which is considered far from systematic and reliable, is
PID regulators. Different sources estimate the share the flaw of this approach. However, this approach could
taken by PID controllers at between 90 and 99%. This is assist in constructing a fuzzy model or an initial version
because of their low cost, inexpensive maintenance, of a fuzzy controller. The second approach supposes an
simplicity of operation, ease of design, and effectiveness application of the knowledge of control engineering and
for most systems [6,17,21]. In practice, it is often being a design of a fuzzy controller in some aspects similar to
integrated into complex control structures in order to the conventional design with the parameter’s choice,
achieve a better control performance. Among those depending on the information of their influence on the
complex control structures, cascade control scheme is controller performance [3-4]. On the other hand best-
commonly used for the purpose of reducing both known industrial process controller is Proportional-
maximum deviation and integral error of disturbance Integral-Derivative (PID) controller because of its simple
responses. The advantages of easy implementation and structure and robust performance in a wide range of
potentially large control performance improvement have operating conditions. The similarity of FLC and PID
led to wide spread applications of cascade control for controllers and there improvement is still being
several decades. It has become a standard application investigated [1].
provided by industrial process controllers [10, 22]. Quang and Negenevitsky have proposed a scheme to
Cascade control systems are constructed by two control tune the PI controller in Cascade loop with Fuzzy Logic.
They have given the simulation result only [20]. Another
Received July 1, 2007; Revised November 1, 2007 and February simulation result was given by Santos and Cruz [18].
15, 2008. Vineet Kumar and K.P.S. Rana are with Dept. However, not many observations have been reported so
Instrumentation and Control Engineering., Netaji Subhas far in the literature on the effect of use of fuzzy logic
Institute of Technology, Delhi University, Sector 3, Dwarka, New controller in the cascade control strategy in real time.
Delhi 110078, India (e-mail: [email protected]); Vandana
The present research work is focused on the
Gupta is with TCS, India.
implementation of
90 INTERNATIONAL JOURNAL OF INTELLIGENT CONTROL AND SYSTEMS, VOL. 13, NO. 2, JUNE

Fuzzy PI + Fuzzy PD Controller in simple feedback and

cascade configuration in real time for liquid-level
process experiment in our laboratory in order to evaluate
the performance of intelligent controller over
conventional PI controller. The level-process is controlled
by manipulating an equal percentage pneumatic control
valve. Due to its equal percentage characteristics it
shows nonlinear behaviour. Fig.3. Block diagram of process
Fuzzy control and Conventional control strategies The system consists of a reservoir tank, which stores the
have been applied to implement level control in the incoming water. This reservoir opens up into an outlet
process control unit as shown in Fig.1. These strategies pipe, which has a valve whose opening can be adjusted
have been successfully implemented in both feedback to vary the disturbance. The pump motor attached to the
and cascade control configurations. In the cascade pipe sucks water from the reservoir when switched on.
control configuration, fuzzy control has been This water travels through the pipes and then reaches the
implemented in only the primary loop. In the secondary control valve, which controls the flow of this water as
loop, only conventional control was used. It has been per its lift, which is in turn controlled by the controller.
observed that the fuzzy controller out perform the The flow of this pipe was gauged by a rotameter which is
conventional controller in both the feedback and cascade installed in the pipe. This pipe, further empty into an
control configurations. overhead tank, in which the level control has been
implemented. The level was sensed by a level sensor
2. FAMILIARIZATION WITH THE PROCESS (pressure type) and the data for the same was acquired
with the help of a data acquisition (DAQ) card.
2.1 Process Control Unit As the incoming water fills in the reservoir, the
desired level (set point) is given as the input to the
Fig.1 & Fig.2 gives a description of the process computer and according to the error in the current level
control unit and Liquid Level Process. The block of the tank (as sensed by the level transmitter) and in the
diagram of process is shown in Fig.3. In this unit the input, the computer generates a control action which was
level in the overhead tank has to be controlled using converted using a DAQ card into an analog signal. This
feedback control and cascade control. is transmitted to the I-P converter, which converts the
current to pneumatic signal (i.e. 4-20mA to 3-15psi). The
pneumatic signal then increases or decreases the opening
of the control valve so as to attain the desired flow in the
pipes, which can eventually give the desired level in the
overhead tank. This is how level is controlled in the
above system using a feedback loop.

2.2 Control Valve

A pneumatic control valve of equal percentage

characteristics was used to regulate the liquid flow-rate.
Fig.1. Process Control Unit The plug of the valve was Globe type and it was
operating in air-to-open mode. The valve was in use for
the last 5 years and experienced wear and tear, rusting,
and accumulation of dust particles. As a result the valve
was not able to get opened to full-scale. The flow
characteristic of pneumatic control valve is shown in Fig.4.
The input-output characteristic of control valve shows
highly nonlinear behavior. Therefore, due to valve the
overall system becomes highly complex and nonlinear.

Fig.2. Liquid-Level Process Setup

 Kumar et al: Real-Time Performance Evaluation of a Fuzzy PI + Fuzzy PD Controller for Liquid-Level Process

Fig.4 Flow characteristic of pneumatic control valve

2.3 Data Acquisition Fig.5. Cascade control in real time (for level control)
4. FUZZY PI + FUZZY PD CONTROLLER
A PCI compatible DAQ card (NI-6035E, 200KS/s,
12Bit) was used to acquire the data and also to generate A conventional PID controller follows the control law:
the control signal. Input and output were synchronized 1
appropriately. A virtual instrument was developed for u (t)  K {e(t)   e(t)dt   de(t)} ……… (1)
this task. PID C  D dt
I

2.4 Transducer where KC is the proportional constant of PID controller,

 is an integral constant,  is a derivative time constant,
I D
Pressure-type level sensor was used to sense the level e(t) is an error signal, e(t)  r(t)  y(t) , r(t) is the
in a tank/reservoir and then convert the sensed signal desired value, y(t) is the output from process and
into equivalent electrical signal (in the range of 4-
20mA). uPID (t) is the output from controller - the action value. Its
Rotameter was used to measure the flow through a discrete approximation is:
pipeline and then convert this flow signal into an
electrical
analog signal (in the range of 4- k e(k)  e(k 1)
u (k )  {e(k )   e(i)T 
20mA). K 1 }, (2)
3. CASCADE CONTROL LOOP
PID C
 i1 S D T
I S
where T the sampling period, k is the step. To implement
S
For controlling level in the overhead tank, using
cascade control, the disturbance caused by the inflow rate the Fuzzy PID controller three inputs e(k) ,  e(k) and
was minimized. The block diagram of cascade control e(k) are required. Increasing the number of input
system is shown in Fig.5.The level transmitter senses the variables causes a rise in the dimension of the rule table
level in the overhead tank and transmits the signal to the and, therefore, in the complexity of the system; this makes
level controller built in the computer. This level its implementation more complicated. For this reason, a
controller computes a set point for the flow loop, which combination of Fuzzy PI + Fuzzy PD controller was
is then fed to the flow controller. The flow controller employed instead of three input Fuzzy PID controller. The
provides an actuating signal to the control valve in a Fuzzy PI + Fuzzy PD controller is a digital controller,
manner so as to nullify the disturbance due to flow which contains a Fuzzy PI & Fuzzy PD control unit’s
before it is able to affect the primary control variable, arrangement [11, 16].
level. This actuating signal is proportional to the error
calculated by the flow controller, which is the difference 4.1 Fuzzy PI Controller
between the computed value of the flow by the level
controller and the actual value of the flow in the pipe A classical PI controller is described by
leading to the overhead tank, measured by the rotameter. 1
The changed flow rate in the
pipe connecting the overhead tank will cause the inner u (t)  K (e(t)  ................................. (3)
loop to respond before the error in the level becomes PI C I  e(t)d
(t)).
evident in the overhead tank. So the purpose of the inner Differentiating equation (3), the following equation is
loop is to detect a disturbance before it becomes evident obtained:
as error in the outer loop.
92 INTERNATIONAL JOURNAL OF INTELLIGENT CONTROL AND SYSTEMS, VOL. 13, NO. 2, JUNE

duPI (t) de(t) 1

 e(t)), Finally, the overall Fuzzy PI + Fuzzy PD controller can
 KC ( be obtained by algebraically summing the equation (4) &
dt dt  I equation (6).
In discrete form, the above equation can be written
as: uPID (k )  uPI (k )  uPD (k ),
[uPI (k)  uPI (k 1)] [e(k)  e(k 1)] 1 KUPI
or u (k)  u u(k)  (k). .................. (7)
PID PI PD
 KC (  e(k )), (1 z1)
TS TS I
1 Fig.6 shows the configuration of Fuzzy PI + Fuzzy PD (in
or, u PI(k )  K C (e(k )  e(k )), feedback mode) [19].
I

or KC
e(k )  e(k 4.4 Fuzzy PI +Fuzzy PD Controller in cascade loop
PI (k ) 
u KC ),

I
In cascade control mode, normally a conventional P or PI
where e(k )  e(k)  e(k 1) , controller is used for the secondary controller (inner
TS controller); because the disturbances arising within the

and secondary loop are corrected by second controller before

uPI (k )  uPI (k )  uPI (k 1) they can affect the value of the primary controller (outer
, controller) output. Any offset caused by P control in the
TS
or TS uPI  uPI (k)  uPI (k 1), secondary loop is not important since we are not
or T u  u (k)  z1u (k ), interested in controlling the output of the secondary
S PI PI PI

or T u (k )  (1 z1 )u process. Therefore, in the present case conventional PI

(k ),
S PI PI
controller was implemented for secondary controller in
or uPI (k ) TS order to get the faster response in secondary controller
u (k),
(1 z1) PI and Fuzzy PI + Fuzzy PD controller for primary

K controller. Fig.7 shows the implementation of Fuzzy PI +
Fuzzy PD
uPI (k ) 
UPI
or u (k ) ................................. (4) controller in cascade control loop [2,7,19].
(1 z1) PI
.
where, KUPI (Gain PI) is the fuzzy control gain.

A physical meaning of the parameters for the fuzzy PI

controller remains the same like for the PI controller (the
controller gain KC and the time integral constant τI) [6,
16, 19].
Fig.6. Fuzzy PI + Fuzzy PD (in feedback mode)
4.2 Fuzzy PD Controller

A classical PD controller is described by:

u (t)  K (e(t) 
de(t)
PD C D ) .................……………….(5)
dt
In discrete form, the above equation can be written
as: Fig.7 Fuzzy PI + Fuzzy PD (Cascade) Configuration

u (k )  [e(k)  e(k 1)]

K (e(k)  ),

PD C D
TS

or uPD (k)  KC (e(k)  De(k)),

or uPD (k )  KC e(k )  KCDe(k ), ......................... (6) Fig.8. Membership function of input and output of FLC

where
e(k)  e(k 1)
e(k )  5. PRE-R EQUISITES FOR DESIGN OF FLC
.
TS
A fuzzy PD controller can be realized as per the above
equation [6, 16, 19]. 4.3 Combination of Fuzzy PI +Fuzzy PD Controller
5.1 Fuzzification Kumar et al: Real-Time Performance Evaluation of a Fuzzy PI + Fuzzy PD Controller for Liquid-Level Process

Input and output variables of the FLC are usually

quantized into sets of classes defined by linguistic
labels such as “positive large”, “positive medium”,
“positive small”, “zero”, “negative small”, “negative
medium”, “negative large”, and so forth. For the
experimentation,
94 INTERNATIONAL JOURNAL OF INTELLIGENT CONTROL AND SYSTEMS, VOL. 13, NO. 2, JUNE

the inputs and outputs were quantized into 5 fuzzy sets, 5.4 Defuzzification
namely:
In the present work center of gravity defuzzification
PL – Positive method (COG) was used to defuzzify the fuzzy sets into
Large PS – a crisp control signal [17].
Positive Small ZE
– Zero 6. REAL TIME RESULTS
NS – Negative Small
NL - Negative Fuzzy PI + Fuzzy PD controller described in last
Large section was implemented in real time in feedback and
cascade control configurations for controlling the liquid-
This quantization was carried out in the range of [-1,1]
for inputs as well as output. The membership functions
for
e(k), e(k) and u PID (k ) were of triangular type and are level in the overhead tank shown in the process control
unit Fig.1. The National Instrument™, USA hardware and
shown in Fig.8 [16]. software (LabVIEW™) were used for real time
implementation of measurement and control. LabVIEW™
5.2 Rule Base (Laboratory Virtual Instrument Engineering Workbench)
is a powerful and flexible instrumentation and analysis
Theoretically, the rule based should be different for software development application created by National
Fuzzy PI controller and Fuzzy PD controller but in order Instrument™, USA.
to reduce the complexity of design and to increase
efficiency, a simple structure of Fuzzy PI + Fuzzy PD
controller was used with a single rule base. A PI rule
6.1 Performance criterion
base was considered because PI controller is generally
more important for steady state performance [8]. For controlling level in the overhead tank of the process
The rule base for this Fuzzy controller can be control unit the main performance evaluation criteria was
imagined to be a two dimensional matrix as summarized taken as peak of overshoot. This is because in the level
in Table 1.The rows represent the various linguistic control system the main thing which has to be taken care
of is the overflowing of the tank. If the overshoot is more
values that change of error e(k ), can take and the tank might overflow. Thus, the main factor which has
columns indicate been considered for evaluating the performance is the
the values of error e(k ) . The entries in this matrix would overshoot. The next priority in the performance criteria
be the control action that has to be taken described in the was given to the settling time i.e. how quickly the level
linguistic terms. The control action has been calculated settled down.
based upon the process reaction curve [3-4].

Table1: Rulebase for Fuzzy Controller

Change of error (de)
NL NS ZE PS PL
NL NL NL NL NS ZE
Error NS NL NL NS ZE PS
(e) ZE NL NS ZE PS PL
PS NS ZE PS PL PL
PL ZE PS PL PL PL

5.3 Fuzzy Inference Engine Fig.9. Fuzzy PI + Fuzzy PD real time response
(feedback)
The basic function of the fuzzy inference engine is to
compute the overall value of the control output variable
based on the individual contribution of each rule in the
rulebase. For the present research work, Mamdani
inference mechanism has been used. The differences in
using the various implication techniques are described in
[5]. It was observed that Mamdani’s technique is the
most suitable for hardware implementation due to simple
min– max structure. The first phase of Mamdani’s
implication involves min-operation since the antecedent Fig.10. Conventional PI real time response (feedback)
pairs in the rule structure are connected by a logical
‘AND’. All the rules are then aggregated using a max-
operation [16]. 6.2 Settings Used
 Kumar et al: Real-Time Performance Evaluation of a Fuzzy PI + Fuzzy PD Controller for Liquid-Level Process

6.4 Results of Cascade Control Configuration

The main settings which had to be done in the system
were of the set-point and the disturbance (this was The various results obtained on implementing the
changed by adjusting the partially open valve shown in fuzzy and conventional controllers, in the cascade loop,
Fig.2.). The settings of these two were kept same for the for controlling the level in the overhead tank of the
whole experiment for carrying out the comparison of the process control unit are shown in Fig.11 & Fig. 12. The
readings. The height of the overhead tank is 80 cms so Fuzzy controllers were implemented only in the primary
the set-point was kept at the middle point i.e. at 40 cms loop of the cascade controllers, while the secondary loop
as the flow verses height characteristics were non-linear had conventional PI controller.
and were linearised at mid-point. The disturbance was set
at 4- full hand turns. 6.4.1. Fuzzy PI +Fuzzy PD Controller Response
Fig.11 shows the real time response from Fuzzy PI
6.3 Results of Feedback Configuration +Fuzzy PD controller in primary loop. The settings of the
parameters for this controller as per Fig. 7 are given below:
The results obtained on implementing the fuzzy and
Primary loop (Fuzzy Controller):
conventional controllers, in the feedback loop, for
Gain e = 1, Gain de = 1, Gain PI = 20, Gain D = 30
controlling the level in the overhead tank of the process
Secondary loop (Conventional PI controller):
control unit are given below.
Kc= 5, i = 0.01
6.3.1. Fuzzy PI +Fuzzy PD Controller Response
Fig. 9 shows the real time response from Fuzzy PI + 6.4.2. Conventional PI Controller Response
Fuzzy PD controller. The settings of the parameters for Fig.12 shows the real time response from
this controller as per Fig. 6 are given below: Conventional PI controller. The settings of the
Gain e = 1, Gain de = 5, Gain PI = parameters for this controller had been calculated using
10, Gain PD = 20, Setpoint = 40 the Zeigler Nichols tuning method. The settings obtained
cms. were:
6.3.2. Conventional PI Controller Response Primary loop (conventional PI controller):
The conventional PID controller was tuned using Kc = 597.5, i = 0.022
Ziegler Nichol’s method. The performance of
Secondary loop (conventional PI controller):
conventional PI controller was best among other
conventional controller. Fig. 10 shows the real time Kc = 5, i = 0.01
response from a conventional PI controller. The settings
of parameters for this controller were calculated using
the relay method.
Kc = 9.6505;  = 0.1528 min; Setpoint = 40cm;
i

6.3.3. Comparison of Fuzzy and Conventional controller

real time results for Level system (Feedback)

The Table 2 compares the results of all the

controllers, Fuzzy and Conventional in the feedback Fig.11. Fuzzy PI + Fuzzy PD response (cascade)
loop. It was observed that the fuzzy controllers
performed better than the conventional PI controller.
Table 2: Comparison of Fuzzy and conventional controllers
real time results (feedback)
Type of Controller Rise Peak Overshoot Settling
Time (sec)(cms) (%) Time (sec)
Fuzzy PI + Fuzzy 169 42.6 6.5 313
PD
Conventional PI 177 47.24 18.1 714

Table 2 shows that Fuzzy PI + Fuzzy PD controller Fig.12. Conventional PI response (cascade)
stands out in performance with the least peak of 42.6 6.4.3. Comparison of Fuzzy and Conventional controller
cms. The settling time has been calculated using the 5% real time results for Level system (Cascade)
tolerance band criteria. The Fuzzy PI + Fuzzy PD
controller has the least settling time of 313 sec. Table 3 compares the results of Fuzzy and
Conventional controllers in cascade loop. It was observed
 96 INTERNATIONAL JOURNAL OF INTELLIGENT CONTROL AND SYSTEMS, VOL. 13, NO. 2, JUNE

that the fuzzy controllers performed better than the [3] C. C. Lee, “Fuzzy Logic in control systems: Fuzzy Logic
Controller – Part I,” IEEE Trans. Syst. Man Cybern., vol. 20, no.
conventional PI controller in cascade loop. The table
2, Mar./Apr. 1990, pp. 404-418.
shows that Fuzzy PI + Fuzzy PD controller stands out in [4] C. C. Lee, “Fuzzy Logic in control systems: Fuzzy Logic
performance according to our chosen criteria, with the Controller – Part II,” IEEE Trans. Syst. Man Cybern., vol. 20, no.
least peak of 41.8 cms i.e. 4.5% overshoot. It’s settling 2, Mar./Apr. 1990, pp. 419-435.
[5] F.S.Smith and Q.Shen, “Selecting Inference and Defuzzification
time has been calculated using the 5% tolerance band
Techniques for Fuzzy Logic Control”, Proc. of UKACC Intern.
criteria. It was observed that the settling time has been Conference on Control, 1–4 Sept. 1998, pp. 54–59.
reduced to 194 sec, which is less than the conventional [6] G. Chen, “Conventional and fuzzy PID controllers: an overview”,
PI in cascade configuration. International Journal of Intelligent and Control Systems, 1, 1996,
pp. 235±246.
Table 3: Comparison of Fuzzy and conventional [7] G. Stephanopoulos, “Chemical Process Control: An Introduction
controllers real time result (cascade) to Theory and Practice”, Prentice-Hall of India, New Delhi, 1984.
Rise Settling [8] H. X. Li and H. B. Gatland, “Conventional Fuzzy Control and Its
Peak Overshoot Enhancement”, IEEE Trans. Syst., Man, Cybern. B, vol. 26, Oct.
Type of Controller Time Time
(%)
(sec) (cms)
1996, pp. 791–797.
(sec)
[9] H. X. Li, “A comparative design and tuning for conventional fuzzy
Fuzzy P I + Fuzzy PD (Primary) control,” IEEE Trans. Syst., Man, Cybern. B, vol. 27, Oct. 1997,
194 41.8 4.5 194
& Conventional PI (Secondary) pp. 884–889.
Conventional PI (Primary) [10] I. Kaya, N. Tan and D. P. Atherton, “Improved Cascade Control
183 42.5 6.25 364
& Conventional PI (Secondary) Structure and Controller Design”, Proceedings of the 44th IEEE
Conference on Decision and Control, and the European Control
Conference 2005 Seville, Spain, December 12-15, 2005, pp 3055-
7. CONCLUSIONS 3060.
[11] J. Carvajal, G. Chen and H. Ogmen, “Fuzzy PID Controller:
Design, Analysis, performance evaluation, and stability analysis”,
Fuzzy PI + Fuzzy PD and conventional controllers Information Sciences, 123, 2000, pp. 249-270.
were successfully implemented in real time, using both [12] J. Zumberge and K.M. Passino, “A case study in intelligent vs.
feedback and cascade control loops. The overall system conventional control for a process control experiment”, Intelligent
Control, 1996., Proceedings of the 1996 IEEE International
performance was realized using the quality hardware for Symposium on , 15-18 Sept. 1996, pp 37 – 42.
the measurement and state of the art software tools like [13] J. Zumberge and K.M. Passino, “A Case Study in Intelligent vs.
LabVIEW™ with associated add-on modules. The Conventional Control for a Process Control Experiment”, Journal
performance of controller in cascade configuration was of Control Engineering Practice, Vol. 6, 1998, No. 9, pp. 1055–
1075.
better than feedback configuration with respect to the [14] K. Passino, “Towards bridging the perceived gap between
performance criteria. From the Table 4, it was observed conventional and intelligent control”, in Intelligent Control:
that the Fuzzy PI + Fuzzy PD controller in cascade Theory and Applications, IEEE Press, 1996, ch. 1, pp. 1–27.
configuration (in primary controller) perform superior Gupta, M.M. and Sinha, N.K. editors.
[15] K. Passino, “Bridging the gap between conventional and
than the conventional PI controller in both the feedback intelligent control”, Special Issue on Intelligent Control, IEEE
and cascade control configuration and fuzzy controller in Control Systems Magazine, vol. 13, June 1993, pp. 12–18.
feedback loop configuration. [16] K. M. Passino and S. Yurkovich, “Fuzzy Control”, Menlo Park,
CA: Addison Wesley Longman, 1998.
Table 4: Comparison of real time results of feedback and [17] K. S. Tang, K. F. Man, G. Chen and S. Kwong, “An Optimal
cascade Fuzzy PID Controller”, IEEE Transactions on Industrial
controller configuration Electronics, vol. 48, no. 4, August 2001, pp. 757-765.
Feedback Cascade [18] M. Santos, J. M. de la Cruz, S. Dormido. "Self-Tuning of Fuzzy
Logic Controllers in Cascade Loops", in Intelligent Systems and
Type Of Controller Tr Peak Mp Ts Tr Peak Mp Ts
Soft Computing for Nuclear Science and Industry, World Scientific,
(sec) (cms) (%) (sec) (sec)(cms) (%) (sec)
1996, pp. 258-264
Fuzzy PI + Fuzzy 169 42.6 6.5 313 194 41.8 4.5 194 [19] P. Pivonka; “Comparative Analysis of Fuzzy PI/PD/PID
PD Controller Based on Classical PID Approach”, Fuzzy Systems,
Conventional PI 177 47.2418.1 714 183 42.5 6.25 364 2002. FUZZ- IEEE'02. Proceedings of the 2002 IEEE International
Tr: rise time, Mp: Maximum overshoot in percentage, Conference, Vol. 1, 12-17 May 2002, pp.541 – 546.
Ts: settling time. [20] Q. P. Ha & M. Negenevitsky, “Cascade PI-Controller with Fuzzy
ACKNOWLEDGEMENT Tuning” , Proceedings of the Sixth IEEE International Conference
on Fuzzy Systems, vol.1, 1-5 Jul 1997, pp 361-366.
The authors are thankful to the management of the Netaji [21] S. Bennett, “Development of the PID controller,” IEEE Contr.
Syst. Mag., vol. 13, Dec. 1993, pp. 58–65.
Subhas Institute of Technology for providing excellent [22] S.Song, L.Xie and Wen-Jim Cai, “Auto-tuning of Cascade Control
experimental facilities in the Advanced Process Control Systems” IEEE Procedings of the 4th world Congress on Intelligent
Lab. Control and Automation, June 10-14, 2002, Shanghai, P.R.China,
pp 3339-3343.
REFERENCES
[1] B.G. Hu, G. K. I. Mann, and R. G. Gosine “A systematic study of
fuzzy PID controllers-function-based evaluation approach”, IEEE
Transactions on Fuzzy Systems, Vol.9, Issue 5, Oct. 2001, pp.699
- 712.
[2] B. A. Ogunnaike and W. H. Ray, “Process Dynamics, Modeling,
and Control”, Oxford University Press, New York, 1994.
 Kumar et al: Real-Time Performance Evaluation of a Fuzzy PI + Fuzzy PD Controller for Liquid-Level Process

Vineet Kumar received M.Sc. degree in K.P.S Rana received his M.Tech. degree from
Physics with Electronics in 1991 from G. B. IIT Delhi and currently pursuing his Doctoral
Pant. Institute of Ag. & Technology, degree in the area of computer based test and
Pantnagar, India and M. Tech. in measurement.
Instrumentation from Regional Engineering He has served Indian Space Research
College, Kurukshetra, India and currently Organization [ISRO] for over seven years. At
pursuing Ph.D. degrees in Intelligent NSIT he is working in the capacity of Assistant
Process Control from Delhi University, Professor since year 2000. He has established
Delhi, India. Instrumentation Laboratory in the department of
He is a Senior Lecturer in the Department of Instrumentation and Control Engineering. His current research
Instrumentation and Control Engineering, Netaji Subhas interest is intelligent instrumentation systems. He has published
Institute of Technology (NSIT), Delhi University, Delhi, India over 22 research papers in international and national
since 2005. He was a lecturer from 2000 to 2005 at NSIT. He conferences and journals.
has also served industry more than 5 years. He has published
more than 17 papers in various international and national
conferences and journals.
He has developed the Advanced Process Control Lab for
research and development with the help of latest hardware and
software from National Instrument™, USA
 98 INTERNATIONAL JOURNAL OF INTELLIGENT CONTROL AND SYSTEMS, VOL. 13, NO. 2, JUNE

View publication stats