You are on page 1of 5

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

net/publication/339097455

PSO Optimized PID Controller Design for the Process Liquid Level Control

Conference Paper · October 2019


DOI: 10.1109/RDCAPE47089.2019.8979108

CITATIONS READS

26 613

3 authors, including:

Adnan Mukhtar
Government College of Engineering and Technology (GCET), Safapora, Ganderbal
8 PUBLICATIONS 41 CITATIONS

SEE PROFILE

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

Estimation of State of charge (SoC) and State of Health (SoH) of Li-Ion batteries in Electric vehicles View project

Optimization of a Robust Controller for the Speed Control of a Separately Excited DC Motor Using Genetic Algorithm View project

All content following this page was uploaded by Adnan Mukhtar on 28 February 2020.

The user has requested enhancement of the downloaded file.


2019 3rd International Conference on Recent Developments in Control, Automation & Power Engineering (RDCAPE)

PSO Optimized PID Controller Design for the


Process Liquid Level Control
1
Adnan Mukhtar,2Vijay Kumar Tayal,3HP. Singh
Department of Electical and Electronics Engineering
Amity University Uttar Pradesh, Noida, India
1
adnan.mukhtar143@gmail.com, 2vktayal@amity.edu, 3hpsingh.2@amity.edu

Abstract—In modern industrial applications, the process terms. The PSO is easy to implement and does not have local
liquid level needs to be maintained at a pre-defined level to maxima/minima and convergence problems.
achieve higher efficiency and final product quality. For
uninterrupted electric power generation, the boiler tank water In this paper, the process liquid level tank has been
level has to be continuously controlled and monitored. The controlled by PID and PSO-PID controllers in MATLAB.
Proportional-Integral-Derivative (PID) is commonly used in such After analyzing the control effect and comparing with PSO-
processes to attain better response. The PID Controllers have PID, the results are found far better than conventional PID.
lesser development costs, superior features and yields in better This work is structured in a systematic way- section II defines
system output. In this paper, the PID gain terms are tuned by the process modeling, PID controller design is presented in
Particle Swarm Optimization (PSO) algorithm. The PSO has section III, section IV illustrates PSO application to design PID
simple concept, implementation is simple. The process liquid level controller, simulation results are discussed in section V and
control of a plant by conventional PID and PSO optimized PID conclusion is presented in section VI.
controllers has been analyzed and quantitatively compared. The
computer simulation results show far better settling time and II. PROCESS MODELING
peak values with PSO optimized PID control, thus ensuring
overall higher efficiency.
The process modeling comprises of basic control scheme,
valve and liquid tank. The basic control scheme for process
Keywords—Matlab, System Model, PID Controller, PSO liquid level is presented in Fig. 1
optimized PID

I. INTRODUCTION
In modern industrial applications, a number of processes
are controlled by various control schemes. The cost effective
PID controllers are one of the best options till date due to their
superior control capabilities. In process control industries,
controlling height of liquid is one of the basic issues faced. The
liquid may be processed in various tanks through pumps. The
chemicals on process liquid level need to be maintained during Fig. 1. Basic control scheme for process liquid level control
various chemical processes. Further due to Interaction between
two tanks, there is a need to control the process liquid level. In A. Valve
power generating plants, for continuous and efficient power
production there is need to monitor and maintain the boiler
water tank level. In many industries, due to non-linearity
present, the conventional PID control scheme is not easy to
implement [1] [2]. Fuzzy control systems are also long in use
for process level control of a plant. However, the non-linearity
present with Fuzzy controller may aggregate the problem [3-6].
Such problems may be resolved by using Artificial Intelligence
technique to optimize the controller parameters for better
output. In literature various Artificial Intelligence techniques
such as Ant colony optimization (ACO), Honey Bee algorithm Fig. 2. Block diagram of valve
(HBA), Teaching Learning algorithm (TLA), Genetic
algorithm (GA) and Particle Swarm Optimization (PSO) have The block diagram of valve is presented in Fig. 2. Integrator
been used for optimal control design problems. GA suffers with saturation is used for controlling the liquid flowing into
from the convergence and multiple maxima/minima problems. the tank.
The Particle Swarm Optimization (PSO) is based on swarm
intelligence is explored in this paper to optimize the PID gain

978-1-7281-2068-3/19/$31.00 ©2019 IEEE 590


2019 3rd International Conference on Recent Developments in Control, Automation & Power Engineering (RDCAPE)

B. Valve

Fig. 4. Block Diagram of PID controller


Fig. 3. Schematic diagram of liquid tank
The block diagram of PID controller is presented in Fig. 4.
The schematic diagram of liquid tank is presented in Fig. PID controller is known by its three gain terms KP, Ki, and Kd
3.The input liquid flow rate into the liquid tank is Qinput and which are proportional, integral and derivative respectively.
the output liquid flow rate is Qoutput. The liquid density is These three gain terms are added to obtain the output of PID.
considered to have a constant value. Routput is the output Feasibility, simple implementation and faster response (by
resistance. HL is the height of the liquid. The liquid flowing minimizing the time constant) are considered to be the main
out may be laminar or nonlinear. advantages of PID controller. Using Matlab/Simulink based
on approximate model PID controller is designed. For the
For the liquid flow to be laminar
desired control response, gain terms (KP, Ki, Kd) are adjusted
ࡴࡸ to their optimum values. Optimization of the gain terms is
Qoutput(t) = (1)
done by hit and trial and by PSO algorithm.

For the liquid flow to be turbulent, IV. THE PARTICLE SWARM OPTIMIZATION (PSO)

Qoutput(t) = ݇ඥ‫ܪ‬௅ (2) PSO is a method of computation with the help of which a
problem is optimized by trying a certain number of iterations.
Where, K is the coefficient of discharge. It is a technique that solves problems which are non-linear and
C. Model Equation non-continuous in nature. The advantages of this technique
are: its concept is simple, easily implemented, control
Liquid flows into the tank from the top and flows out from the parameters are robust, and high computational efficiency. This
tank at the bottom through inlet and outlet valves respectively.
technique is explored in this paper to optimize the gain terms
The systems model is attained by relating the input flow with
of PID controller .
the output flow in the liquid tank. Using a balance of input and
The particles update their velocities given by equation (8)-
output flow equations on liquid tank
ഥ൫‫۾‬ሺ‫ܜ‬ሻ െ ‫܆‬ሺ‫ܜ܂‬ሻ൯ ൅ ‫܉‬
‫ڂ‬ሺ‫ ܜ‬൅ ૚ሻ ൌ ‫܃܅‬ሺ‫ܜ‬ሻ ൅ ‫܉‬ ഥሺ۵ሺ‫ܜ‬ሻ െ ‫܆‬ሺ‫ܜ‬ሻሻ (8)
Qinput(t) - Qoutput(t) =‫ܣ‬஼ ݀‫ܪ‬௅ ሺ‫ݐ‬ሻ ോ ݀‫ݐ‬ (3)
The particles update their positions by equation (9)-
Where AC is cross sectional area of liquid tank. The output ‫ܡ‬ሺ‫ ܜ‬൅ ૚ሻ ൌ ‫ܡ‬ሺ‫ܜ‬ሻ ൅ ‫܃‬ሺ‫ ܜ‬൅ ૚ሻ (9)
flow is assumed to be turbulent in nature. So
Where P(t) and G(t) are best preceding positions known by its
Qoutput(t) = ݇ඥ‫ܪ‬௅ (4) neighbors. W is a constant, ƒത ‫[ א‬0 a].

Now equation (3) can be written as The PSO parameters used in this work are depicted in Table I

Qinput(t) - ݇ඥ‫ܪ‬௅ =‫ܣ‬஼ ݀‫ܪ‬௅ ሺ‫ݐ‬ሻ ോ ݀‫ݐ‬ (5) TABLE I. PSO PARAMETERS

S.NO PSO Parameters


Qinput(t) =‫ܣ‬஼ ݀‫ܪ‬௅ ሺ‫ݐ‬ሻ ോ ݀‫ ݐ‬+ ݇ඥ‫ܪ‬௅ (6)
01 Initial weight 0.9
For the modeling of system, transfer function [9] is given by
equation (7)- 02 Final weight 0.4

૙Ǥ૚૞૜૜ 03 Acceleration constants (a1, a2) 1.4455, 1.4455


GP(S) = ࢋି૙Ǥ૚ૢࡿ (7)
‫܁‬
04 Size of population 40
For PID controller tuning and simulation, transfer function is
05 Maximum number of iterations 50
used.

III. DESIGN OF PID CONTROLLER

591
2019 3rd International Conference on Recent Developments in Control, Automation & Power Engineering (RDCAPE)

simulations are carried out. The desired process liquid level is


presented in Fig. 6. The Simulink models of the process
without and with controller are presented in Fig. 7 and Fig. 8.

Fig. 6. Reference Water Level of the control process

Fig. 5. PSO flow chart

The PSO flow chart is presented in Fig. 5.


The Performance Index (J) [10], i,e Integral of time multiplied
by absolute error (IATE) is chosen as an objective function
shown in equation (10)-

Fig. 7. Simulink Model of a Process without Controller
‫ ܬ‬ൌ  ‫׬‬଴ ‫ݐ‬ȁ݁ሺ‫ݐ‬ሻȁ݀‫ݐ‬ (10)

The PID controller parameters (KP, Ki, Kd) obtained for


conventional controller (by hit and trial) and PSO algorithm are
shown in Table II.

TABLE II. PID CONTROLLER PARAMETERS


S.NO Gain of Controllers Conventional PID PSO-PID
01 Kp 11 9.9538

02 Ki 05 0.040

03 Kd 0.0001 0.1472

V. SIMULATION RESULTS
In this work, the level of process has been controlled by using
PID and PSO-PID controllers. By the use of Matlab/Simulink
Fig. 8. Simulink Model of a Process with Controller

592
2019 3rd International Conference on Recent Developments in Control, Automation & Power Engineering (RDCAPE)

The process liquid level with conventional PID and PSO-PID VI. CONCLUSION
controllers are presented in Fig. 9 and Fig. 10 respectively. The liquid level in a boiler of power producing station needs
The process liquid level in two cases is summarized in Table to be constant for uninterrupted and efficient power
III. production. In this chapter, to attain better results, PSO
optimized PID and conventional PID are designed to control
the process liquid level. The results shown by the PSO
optimized PID over conventional PID are far better in terms of
peak value and settling time. The quantitative comparison of
time domain specifications viz.-settling time and peak value
with PID and PSO-PID indicates much superior performance
by PSO-PID controller.

REFRENCES
[1] Maziyah Mat Noh, Muhammad Sharfi Najib and Nurhanim Saadah
Abdullah, “Simulator of Water Tank Level Control System Using PID-
Controller”, 3rd IASME/WSEAS Int. Conference on Water Resources,
Hydraulics and Hydrology, University of Cambridge, 2008.
[2] Isa, I.S., Meng, B.C.C., Saad, Z., Fauzi, and N.A., “Comparative study
of PID controlled modes on automatic water level measurement system”,
Fig. 9. Process Liquid Level with conventional PID Controller 7th International Colloquium on Signal Processing and its Applications,
Penang, 2011.
[3] Qi Li, Yanjun Fang, Jizhong Song and Ji Wang, “The Application of
Fuzzy Control in Liquid Level System”, International Conference on
Measuring Technology and Mechatronics Automation, 2010.
[4] George Stephanopoulos, “Chemical Process and Control’’.
[5] Wei-Der Chang, “Two liquid-levels system control based on using PI
controllers”, International Conference on Information Security and
Intelligent Control, 2012.
[6] Maruthai Suresh, Gunna Jeersamy Srinivasan and Ranganathan Rani
Hemamalini, “Integrated Fuzzy Logic Based Intelligent Control of Three
Tank System,” vol. VI, Serbian Journal of Electrical Engineering, pp.
1–14, 2009.
[7] Passino, K.M. & Yurkovich, S., “Fuzzy Control”, Addison Wesley,
Menlo Park, 1998.
[8] Jang, J.S., Sun, C.T. & Mizutani, E. “Neuro-Fuzzy and Soft
Computing”, Prentice Hall, New Jersey, 1997.
[9] Jyothish KumarS.Y, “Design and Implementation of Fuzzy Controller
on Embedded Computer for Water Level Control”, International Journal
Fig. 10. Process Liquid Level with PSO-PID Controller of Scientific & Engineering Research, volume 4, issue 5, May-2013.
[10] Vijay Kumar Tayal and Jagdeep Singh Lather, “PSO based Robust
Fuzzy Power System Stabilizer Design for Single Machine Infinite Bus
System”, Annual IEEE India Conference (INDICON), 2015.
TABLE III. COMPARISON BETWEEN CONVENTIONAL PID AND [11] Deepa Shivshant Bhandare, N. R. Kulkarni, “Performances Evaluation
PSO-PID and comparison of PID controller and Fuzzy Logic controller for Process
S.NO liquid level control”, 15th International Conference on Control,
Specifications Conventional PID PSO-PID
Automation and Systems ( lCCAS 2015).
01 Peak value 15 sec 07 sec [12] S.Sundaram, ‘Process dynamics and control’, Cengage Learning, India.
02 Settling time 14.8 PU 13.01 PU [13] Li LIANG, “The application of fuzzy PID controller in coupled-tank
liquid-level control system”, International Conference on Electronics,
03 Steady state error 0 0 Communications and Control (ICECC), 2011.
[14] Aiswarya Lakshmi Sasidharan Nair, S Anitha Janet Mary and J Arul
Linsely, “Modeling and Control of Level Control Process- A
Comparative Study”, International Conference on Innovations in Power
From Fig. 9-10 and Table III, the steady state error obtained and Advanced Computing Technologies.
are zero with conventional and PSO-PID. However, the peak
value and settling time with conventional PID controller are
14.8 pu and 15sec respectively. Whereas, these values with
PSO-PID are 13.1 pu and 07 sec respectively. Thus, higher
values of settling time and peak value obtained with
conventional PID as compared to PSO optimized PID.

593

View publication stats

You might also like