Beruflich Dokumente
Kultur Dokumente
)L[HG6WUXFWXUH+&RQWURORI&RXSOH7DQN6\VWHP
DQG$QWL,QWHJUDO:LQGXS3,'&RQWURO6WUDWHJ\IRU
$FWXDWRU6DWXUDWLRQ
Muhammad Zia Ur RDKPDQ*1, Inam Ul Hassan SKDLNK2, Ahsan AOL3, Nasir AKPDG4
Department of Mechatronics and Control Engineering, University of Engineering and Technology Lahore,
Faisalabad Campus, Pakistan
Department of Electrical Engineering, University of Engineering and Technology Taxila, Pakistan
zia_rahman115@yahoo.com *(Corresponding Author)
Abstract: This paper considered two main issues. First is Classical PID and Conventional ܪஶ synthesis. The proposed
about the complete mathematical modelling and Fixed-Structure controller is a linear robust controller and it overcomes these
ஶ robust controller design for couple tank system in frequency constraints. And proposed controller can take any form such
domain. This system is nonlinear and classical Proportional as simple gain, PI, PD, PID controller, complex lead lag and
Integral Derivative (PID) controller may not give appropriate observer based controller. And parameters of the proposed
performance. Although classical PID controllers are more
controller are tuned using non-smooth ܪஶ optimization in
practical due their simplicity and linearity but they have some
limitations. Especially in case of nonlinear system or small frequency domain. These controllers are easy to implement
variations in parameters, they may not give desired transient practically because are robust linear controllers.
specification. The proposed controller is a robust controller The second part of this paper deals with the major
which contains a fixed structure and fixed order, and its nonlinearity in the form of actuator saturation that may
parameters are optimized using non smooth ࡴஶ optimization in occur in couple tank systems. This nonlinearity is known
frequency domain. Also Fixed-Structure ࡴஶ controller also gives as the Integral windup phenomena and it may lead to
satisfied performance in case of parameters variation. Couple controller in an unstable mode. And the whole system may
tank system is widely used in industry. For example its most be destroyed due to actuator saturation. To overcome this
important applications are water purification, effluent treatment,
problem Anti-integral windup PID control strategy is
dairy filtration, chemical processing and also this system is used
in pharmaceutical industry. Actuator saturation is a major proposed. The proposed Anti-integral windup PID reduces
nonlinearity that may exist in couple tank systems. And this the integral error and keeps the controller in stable mode. In
nonlinearity may lead to performance degradation or instability the last of this paper comparison and conclusion is also given.
of controller and whole system. This problem is known as
integral windup and to deal this problem, an Anti-Integral ,,.0DWKHPDWLFDO0RGHOOLQJ
Windup PID control strategy is introduced in the second part of
$ 'HVFULSWLRQRIWKHV\VWHP
this paper.
A couple tank system which is widely used in industries, it
Keywords: Couple tank system, Classical PID, Non consists of two tanks, orifice valve B to couple tank1 and tank
smooth ࡴஶ optimization, Fixed-Structure ࡴஶ controller, 2, electrical valve for input flow in tank 1. The typical
Integral windup, Anti-integral windup. constraint on maximum opening of electrical valve may occur
in couple tank systems. And the plant input or control input is
, ,QWURGXFWLRQ Qi , which is pump flow rate through electrical valve. A couple
Couple tank systems are widely used in process industries. tank system is shown below in )ig 1 and description of
For example these are used for the purpose of water treatment, symbols is given in 7able ,.
paper making and in petro chemical etc. Liquid flow and
liquid level control in the tanks is the main problem. Many
researchers in [1-3] have worked to investigate this problem in
multi tanks. Researchers have also applied many techniques
to deal this nonlinear system, such as fuzzy control [4]. The
proposed control technique is advanced control technique and
this is yet not investigated on couple tank systems. The
proposed technique is based on ܪஶ theory [5]. But main
advantage of fixed structure ܪஶ controller over conventional
synthesis is that, it could be given fixed order and fixed
structure. The typical design requirements [6] for Single Input
Single Output (SISO) and Multi Input Multi Output (MIMO)
systems such as robust stability, control bandwidth,
disturbance rejection and response speed are constraints on
Figure 1 Industrial Couple Tank System
ൌ ୢ ᩞሺሻ
Table , Description of symbols Where
ୢ =discharge coefficient, ᩞ=cross section area of orifice
Symbols Description
After applying the Bernoulli theorem, equations (1)&(2) will
Qi Flow Rate (control input)
become
H1 Liquid Level in Tank 1 ଵ
୧ െ ୢଵ ȽΌඥʹሺΌ െ ሻሻ ൌ ሺሻ
H2 Liquid Level in Tank 2
Qc Flow Rate Through Orifice valve C And
ଵ
Qb Flow rate through Orifice valve B ୢଵ ȽΌඥʹሺΌ െ ሻሻ െ ୢଶ Ƚඥʹ ଶ ൌ ሺͺሻ
A Cross section area of Both Tanks
ଶ &$3HUWXUEHG/LQHDUL]HG0DWKHPDWLFDO0RGHO
ୠ െ ୡ ൌ ሺʹሻ
We assume that liquid level in tank1 (H1), liquid level in thank
Now by using Bernoulli theorem above dynamic equations is 2 (H2), flow rate Qi and flow rate Qc are at their steady state
converted into nonlinear equations. Bernoulli’s theorem levels. And small variation in all these four parameters will
relates the pressure, elevation and velocity in moving liquid lead above nonlinear equations (9) & (10) into following form.
fluid the flow is steady, where the viscosity and
compressibility are neglected. Bernoulli fluid flow equation is ሺΌ Όሻ
given below. ୧ ୧ െ ჴΌටଵ Όሻ െ ሺ ଶ ሻ ൌ
;
ൌ
ሺ͵ሻ
ʹ ᛄ
where (11)
ൌ
ሺ ଶ ሻ
ൌ
ȽΌඥሺଵ Όሻ െ ሺଶ ଶ ሻ െ Ƚඥଶ ଶ ൌ
ൌ
ൌ
ᛄ ൌ Now subtracting nonlinear equations (9) & (10) from
Now apply this Bernoulli equation in flowing fluid, we have perturbed equations and simplifying we get these equations.
Where And
Ό ് ͲǡΌ ൌ Ͳǡ ൌ Ͳ ȽΌ ȽΌ Ƚ ሺሻ
Ό െ െ ൌ ሺͳ͵ሻ
Solving above equation (4) we get ʹඥଵ െ ʹඥଵ െ ʹඥ
ൌ ඥʹሺΌ െ ሻሻሺͷሻ
Ό Ό మ
And flow discharge equation is following Let Ό ൌ , ൌ
ଶඥୌభ ିୌ ଶඥୌభ ିୌሻ ଶඥୌ
Linear Time Invariant (LTI) state space model for couple tank ,,, &RQWUROOHUVGHVLJQIRU&RXSOH7DQN6\VWHP
system is given below and 7able ,, represents the parameters As LTI perturbed model is derived, now the main challenge
is the designing of a controller which must give appropriate
and their values for couple tank system. transient response and track the reference input. Two kinds of
݄ሶଵ െܭଵ ିଵ controllers are designed and implemented in our paper. These
ቈ = ቈ ܭଵ ଵ
൨ + ቂ ܣቃ ݍ (14) are Fixed-Structure ܪஶ controller and classical PID. The
݄ሶଶ െܭଵ െ ܭଶ ଶ Ͳ control variable is the flow rate into the Tank1 and measured
variable is the liquid level in Tank2. The performance of both
controllers is compared based on tracking of reference input
Table,, Parameters and their values for couple tank system
signal and transient specifications such as settling time, rise
Name Expression Value time, steady state error and overshoot.
Cross section areas A1 & A2 4m2 & 4m2
$&ODVVLFDO3,'&RQWUROOHU'HVLJQ
of Tank1 & Tank2 The classical PIDs are simple and have low cast so that these
Discharge Cd1&Cd2 1 & 0.5 are used extremely everywhere in the world, in the process
coefficients industries. These controllers sometime become less effective
for those dynamic models, which contains nonlinearity
Orifice valve’s a1&a2 1cm2 & 0.8cm2 inherently. And they do not give desired specifications in
cross section areas terms of transient response and desired reference tracking.
The parameters of classical PIDs are optimized. The PID
gravitational G 9.8m/s2
controller is design using the technique pole zero placement.
acceleration The perturbed LTI model of couple tank system is used to
Sensor Ideal 1 design the controller. The time and frequency domain
Offset Liquid H1 & H2 0.5m &0.2m equations of classical PID are given below.
Levels ୲ ୢ
ሺሻ ൌ ୮ ሺሻ ୧ ሺɒሻ ɒ ୢ ሺሻ (16)
ୢ୲
0.6
0.4
0.2
0
0 2 4 6 8 10 12 14 16 18 20
time(s)
Figure 4(a) Step Input Tracking
Figure 3 Structured synthesis in standard form
Robust Control theory [10] mentions that either SISO or 2.5
PID Control
MIMO systems can be arranged in above standard form. Reference liquid level in Tank2 (m)
Where all control elements are placed in C(s) block and Fixed-Structre Hinfinity Control
2
remaining parts of block diagrams are placed in P(s) block.
And the w is comprised of disturbances and commanded
inputs and z is consisted of error signal. 1.5
0.5
From w to z the closed loop system is given by LFT [24]:
ܶ௪௭ (s) = ܨ (P,C) := ܲଵଵ + ܲଵଶ C (I - ܲଶଶ Cሻିଵ ܲଶଵ (18) 0
ౚ ୱ
୨ (s) = ୮ + + (19) Figure 4(b) Stair Sequence Input Tracking
ୱ ୱାଵ
Where 4.5
PID Control
୮ is proportional gain, ୧ is integral gain, is 1st order filter 4
Reference liquid level in Tank2 (m)
Fixed-Structre Hinfinity Control
coefficient and ୢ is derivative coefficient.
The critical design requirements such as disturbance rejection, 3.5
,9 ,QWHJUDO:LQGXS3KHQRPHQDLQ&RXSOH7DQNV 1
Reference liquid level in Tank2 (m)
PID Controller with Actuator Saturation
0.9
The second part of this paper starts now. Integral windup
phenomena are produced in case of actuator saturation. In case 0.8
electrical valve, which provides input flow into Tank1, is the 0.4
major nonlinearity. The block diagram given next represents
the integral windup phenomena in couple tank systems. 0.3
0.2
0.1
0
0 2 4 6 8 10 12 14 16 18 20
time(s)
9. $QWLLQWHJUDOZLQGXS3,'&RQWURO6WUDWHJ\
3
30
2 Control Effort with Anti-integral windup PID Control strategy
20
1
10
0
0 2 4 6 8 10 12 14 16 18 20
Time(s) 0
0 2 4 6 8 10 12 14 16 18 20
Volumetric Flow Rates (cm^3/sec)
80 30
Volumetric Flow Rate (cm^3/sec)
Control Effort Plant input with Anti-integral windup PID control strategy
20
60
10
40
0
0 2 4 6 8 10 12 14 16 18 20
20
6
0
0 2 4 6 8 10 12 14 16 18 20 4
Saturated Control Effort
time(s)
2
1.4
Reference liquid level in Tank2 (m) The other main advantage of Fixed- Structure ܪஶ controller is
Anti-integral windup PID controller
its fixed order and fixed structure. The order of proposed
1.2
controller is fixed as compare to conventional ܪஶ synthesis,
where in case of conventional ܪஶ controller order is increased
1 due to complex weighting transfer functions. The proposed
controller is the robust linear controller and its parameters are
Liquid Level in Tank2 (m)