ICME07-AM-62
ICME07-AM-62
ICME07-AM-62
ABSTRACT
Liquid tank systems play important role in industrial application such as in food processing, beverage, dairy
filtration, effluent treatment, pharmaceutical industry, water purification system, industrial chemical
processing and spray coating. A typical situation is one that requires fluid to be supplied to a chemical
reactor at a constant rate. An upper tank can be used for filtering the variations in the upstream supply flow.
Feedback control system such as classical PID controller based on the linear model is widely used in
controlling industrial liquid level control application. However aggressive performance requirements may
not be achievable with the PID controller due to the presence of nonlinear dynamics inherent in the liquid
control system and system parameter variations caused by for example corrosive build-up in liquid level
systems creates variation of cross section areas of the tank and discharge orifice. One of nonlinearity which
exists in industrial coupled tank system is actuator saturation. It is well known that integrator windup which
may cause instability and/or performance degradation. To solve this problem, in this paper, a robust
Anti-windup PID controller is designed and implemented for level control of tank system. The
experimental results show that the proposed robust Anti-windup PID controller effectively reduces for the
effect of actuator saturation
© ICME2007 2 AM-62
anti-windup system is used instead of pure PID The proposed controller is developed in Simulink using
compensator. The structure of the tracking Anti-windup its blocks, and then it is built so that C code is generated,
PID controller is illustrated in Fig. 3 where KT is called compiled and finally a real-time executable code is
tracking gain [7]. Once PID controller output U(s) generated and downloaded to the Target Computer.
exceeds the actuator limits, a feedback signal will be The Target Computer is another personal computer
generated from the difference of the saturated and the which is booted using xPC boot floppy disk that loads the
unsaturated signals. This signal is used to reduce the xPC Target real-time kernel. Subsequently, the generated
integrator input. Mathematically, the output of the real-time executable code is downloaded to the Target
Anti-windup PID controller is Computer via selected communication protocol without
writing any low-level code. The connection between the
⎡ ⎤ Host Computer and Target Computer is accomplished
U ( s ) = ⎢ K p + i + K d s ⎥ E ( s ) − T [U ( s ) − U s ( s )] (6)
K K
either through serial (RS-232) or network (TCP/IP)
⎣ s ⎦ s
communications. The communication interface have to
where KT, U(s) and Us(s) are tracking anti-windup gain,
be defined during xPC setup process in the MATLAB
control output and saturated control output respectively.
since the communication protocol definition is required
To set the value of KT, a rule of thumb point out that
in creating the xPC boot floppy disk for the Target
KT=Ki may give a good results but higher values may
Computer. In the proposed system, serial
give a further improvement in performance [9].
communications is used since it is inexpensive, easy to
Therefore, the performances of the tracking anti-windup
install and requires only a cable for connecting serial
system depend on selecting the suitable value of KT.
ports of the Host Computer and Target Computer.
4. EXPERIMENTAL SETUP
5. CONTROLLER DESIGN
In order to examine the effectiveness of the proposed
The proposed Anti-windup PID controller has to be
controllers, a series of experiment is done using the
designed their parameters to achieve a specific desired
lab-scale industrial coupled system. The experimental
performance of the level control system. In order to
setup is shown in Fig. 4. The proposed robust
realize fast response, the proposed robust Anti-windup
Anti-windup PID control algorithm is implemented on a
PID controller is designed so that the settling time of
personal computer and are operated with 1 ms sampling
closed-loop system is less than 500 sec. Therefore,
time. The MathWork's MATLAB/Simulink is used for
according to Eq. (6), the closed-loop system should have
real-time controller implementation through RTW and
the natural frequency larger than 0.001143. Then the
xPC Target. The RTW environment provides a real-time
natural frequency of 0.05 is selected. As a result, the
operation using personal computers and multifunction
desired optimal closed-loop transfer function is
I/O boards. However, the use of RTW still requires the
1.25 x10 −4
development of custom interface programs for correct T (s) = 3 (7)
communication with multifunction I/O boards. In order s + 8.75 x10 −2 s 2 + 5.38 x10 −3 s + 1.25 x10 −4
to overcome this problem, xPC Target is included in the The three parameters of the PID controller has to be
software configuration. designed is obtained by direct comparison between Eqs.
(2) and (7). As the result, the PID controller parameters
Simulink & xPC are as follows
Target K p = 7.50 , (8.a)
ADC RS232 K i = 0.181 , (8.b)
K d = 71.7 . (8.c)
Target PC Host PC
Furthermore, in order to make the closed-loop system
transfer function has no any zeros, the following
Sensor 1 Sensor 2
pre-filter transfer function is obtained also by comparing
Tank 1 Tank 2
Eqs. (2) and (7):
2.52 x10 −3
G f (s) = 2 . (9)
s + 1.05 x10 −1 s + 2.52 x10 −3
h1
h2 Finally, in order to suppress the effect of actuator
saturation, the tracking anti-windup gain KT equals to Ki
is used.
Driver
Pump 6. RESULTS
Fig 4: Setup of real-time controller implementation The designed robust Anti-windup PID controller is
tested in the real-time experiment. The performances of
The xPC Target software supports and provides the proposed robust anti-windup PID controller are
built-in drivers for many industry standard DAQ card compared with those of the PID control system. Figs. 4, 5
including the PCI-6024E DAQ card by National and 6 show the responses of the PID controllers when the
Instrument which is used in the prototype of the 5 cm, 10 cm and 15 cm step input references were used
automatic gantry crane system. A personal computer is respectively. The detailed performances are shown in
used as a Host Computer. Windows operating system and Table 2. Here, the performances of level control system
other required software are running in the Host Computer. are evaluated based on overshoot, settling and error at
© ICME2007 3 AM-62
500 sec. a larger overshoot of the system controlled by only PID
6 controller. As shown in the figures and Table 2, the
proposed controller significantly reduced the overshoot
Level h2, cm
8
confirm that the proposed controller is effective since it
give a good performance even if the actuator saturates.
6
4 PID 8. REFERENCES
2 PID + Anti-windup
1. Visioli, A., 2004, “A new design for a PID plus
0
0 100 200 300 400 500
feedforward controller”, Journal of Process Control,
Time, s 14: 457–463.
Fig 5: Experimental responses to 10 cm step input 2. Tan, K.K., Huang, S. and Ferdous, R., 2002,
“Robust self-tuning PID controller for nonlinear
20 systems”, Journal of Process Control, 12: 753–761.
3. Almutairi, N.B. and Zribi, M., 2006, “Sliding mode
Level h2, cm
15
control of coupled tanks”, Mechatronics, 16:
10 427-441.
5
PID 4. Pan, H., Wong, H., Kapila, V. and de Queiroz, M.S.,
PID + Anti-windup 2005, “Experimental validation of a nonlinear
0 backstepping liquid level controller for a state
0 100 200 300 400 500 coupled two tank system”, Control Engineering
Time, s
Practice, 13:27–40.
Fig 6: Experimental responses to 15 cm step input
5. Dorf, R. and Bishop, R., 2007, Modern Control
Table 2: Performance comparisons System, 11th Ed, Prentice Hall, New Jersey.
6. Wahyudi, Fadhil, M. and Shazri, 2007, “Modeling
Reference Performance Anti-windup PID and Parameters Identification of a Coupled
PID Industrial Tank System”, Proc. National
Conference on Software Engineering and Computer
5 cm Overshoot 8.31% 8.31 % Systems (NaCSES’07).
Settling time 500 sec 498 sec
Error 0.198 cm 0.091 cm 7. Bohn C. and D.P. Atherton, 1994. “A SIMULINK
10 cm Overshoot (%) 4.03% 11.6% Package for Comparative Studies of PID
Settling time 434 sec 500 sec Anti-windup Strategies”, Proc. of IEEE/IFAC Joint
Error 0.186 cm 0.248 cm Symposium on Computer-Aided Control System
15 cm Overshoot (%) 2.60% 33.6% Design, 447-452.
Settling time 302 sec 393 sec
Error 0.065 cm -0.152 cm
© ICME2007 4 AM-62