Académique Documents
Professionnel Documents
Culture Documents
Abstract: - This paper presents graphic-based simulation and optimal PI controller design of a static var
compensator (SVC) using MATLAB’s SIMULINK. Utilization of MATLAB software simplifies problem
solving complexity and also reduces working time. In this paper, an optimal PI controller design using genetic
algorithms for the SVC was demonstrated. From which satisfactory controller design results, tests against
power and three-phase fault disturbances were carried out. Also, comparisons between those results obtained
by i) SVC with the PI controller and by ii) using power system stabilizer were examined to evaluate its
effectiveness.
Key-Words: - PI Controller, Genetic Algorithms, Static Var Compensator, Power System Stabilizer
X d Σ′ = X d ′ + X L (9)
X qΣ = X q + X L (10)
X dΣ = X d + X L (11)
′
X dΣ Eq ( X d − X d ′ )Vb cos δ Vtref
Eq = − (15) Vtact
′ ′ KP +
KI
X dΣ CX dΣ s
∆ω ∆δ
X qVb sin δ
1 ω0
2Hs + D s
Vtd = (16) KV K SVC
CX qΣ K1 TSVC s + 1
′
X L Eq Vb X d ′ cos δ
K2 K4 K5
Limiter
No
Measured input Apply GA operators:
Limiter Selection, crossover and mutation
TOOLBOX) is employed to generate a set of initial Variable ranges 0.1 ≤ KP, KI, ≤ 150
random parameters. With the searching process, the Objective function (discretized version):
parameters are adjusted to give response best fitting N
close to the desired response. To perform the
searching properly, its objective function is the key.
Minimize ∑ ti ∆ω(i)
i =1
In this paper, the objective function is defined in
(18).
tsim
∫ t ∆ω(t ) dt (18)
0
…, where
y desired is the desired response
ysimulated is the simulated response
By using MATLAB’s SIMULINK, the control Fig. 8. Control block diagram of the SMIB with
block diagram of each test case can be formed SVC-PI controller formulated by using
graphically. Although this model can be written in MATLAB’s SIMULINK
other programming codes such as C/C++,
FROTRAN, or even MATLAB programming codes, By using MATLAB’s SIMULINK, the simulation
these programs are usually complex and difficult to can be done easily. However, due to its
debug. In addition, it takes very long editing time to simplification, several adjustments during its
modify the programming codes in order to perform a numerical computing procedure cannot be changed
different task. In this paper, a control model for each by users. One of frequent errors found by using the
case is presented in Figs. 7 and 8 for the SMIB SIMULINK is the mismatch of time stepping. To
without and with the SVC, respectively. guarantee accuracy and stability of simulated results,
In this paper, the PI parameter tuning was variable time stepping is applied to most of
performed by using the GAs. The success of the SIMULINK numerical computation block, for
tuning is associated with the GA’s parameter setting. example the ODE solver. Different value of KP and
The following is a set of parameters used in this KI leads to a different number of sampling after the
research. simulation. For example, given that the simulation
time span is set to 0.0 – 1.0 second. If the KP and KI
GA’s parameter setting are 1.0 and 10.0 respectively, the simulation
Population size 20 produces a 1000×1 array for variable x. By adjusting
Crossover probability 80% the value of the KP and KI to 2.5 and 5.0
Mutation probability 1.0% respectively, the variable x becomes a 1150×1 array
Maximum generation counted 1000 for this case. This results in computational
Maximum generation stalled 50 inaccuracy of the objective function. To play with
this inaccuracy, linear interpolation [19] of every
Proceedings of the 7th WSEAS International Conference on Power Systems, Beijing, China, September 15-17, 2007 34
simulated output is required. If a 1000×1 array of To evaluate the effectiveness of the design
variable x is selected, all x obtained by the algorithm, tests against i) power and ii) three-phase
simulation must be interpolated to give a fault disturbances are situated. The obtained results
corresponding value at fixed 1000 positions of the of the power disturbance case are shown in Figs. 11
simulated time span. Fig. 9 describes this and 12 for the generator speed and the generator
interpolation. The circles on the graph indicate fixed terminal voltage variations, respectively. In the same
positions of the interpolation, while the triangles manner, Figs. 13 and 14 also present the simulated
present points obtained by simulating the results of the three-phase fault disturbance case.
SIMULINK model. Using the linear interpolation, As a result, responses due to some disturbances
the squares estimating the simulated outputs at given given in this paper are stabilized and fast when the
positions can be calculated. optimal PI parameter with linear interpolation in
order to solve a problem of computational
inaccuracy is taken into account. Furthermore, it
gives the best response compared with the others.
-4
x 10
6
PSS Case
4 NI SVC Case
IP SVC Case
-2
-4
1.002
0.998
NI case denotes non-interpolation case
0.997
IP case denotes interpolation case 0.996
0.995
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time (s)
0.01
0.005
-0.005
-0.01
-0.015
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time (s)
Fig. 10. Convergence of the PI parameter tuning
using MATLAB’s GADS TOOLBOX for the first Fig. 13. Generator speed variation due to the
100 consecutive generation three-phase fault disturbance test case
Proceedings of the 7th WSEAS International Conference on Power Systems, Beijing, China, September 15-17, 2007 35
1.3
on Power Delivery, Vol. 12, No. 2, 1997, pp.
1.2
PSS Case 941-946.
NI SVC Case
1.1
IP SVC Case [5] H.F. Wang, Phillips-Heffron Model of Power
Systems Installed with STATCOM and
Gen. Terminal Voltage (p.u.)
1
Applications, IEE Proceedings: Generation,
0.9
Transmission and Distribution, Vol. 146, No. 5,
0.8
1999, pp. 521-527.
0.7
[6] Y.P. Wang, D.R. Hur, H.H. Chung, N.R.
0.6
Watson, J. Arrillaga, S.S. Matair, A genetic
0.5
algorithms approach to design an optimal PI
0.4
controller for static VAr compensator, Int. Conf.
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 Power System Technology (PowerCon 2000),
Time (s)
Vol. 3, 4–7 December 2000, pp. 1557-1562.
Fig. 14. Generator terminal voltage variation due to [7] M.A. Abido, Y.L. Abdel-Magid, Investigation of
the three-phase fault disturbance test case the Effectiveness of Exicitation and FACTS-
Based Controllers on Power System Stability
5 Conclusion Enhancement, The 6th Saudi Engineering
This paper illustrates the utilization of MATLAB’s Conference, Vol. 4, December 2002, pp. 281-
SIMULINK to optimal PI controller design for a 296.
static var compensator, which is connected to an [8] M.L. Kothari, J.C.Patra, Design of Statcom
infinite bus power system. The proposed design Controllers with Energy Storage System Using
includes linear interpolation of every simulated GEA, The 37th Annual North American Power
output to distribute all state variables uniformly over Symposium, 23-25 October 2005, pp. 260-266.
all fixed points of calculation. The computational ®
[9] The MathWorks Inc., MATLAB The language
inaccuracy can be eliminated. In addition, the of technical computing, CD-ROM Manual, 2004.
simulation results confirm that the optimal PI ®
controller design using the GAs together with linear [10] The MathWorks Inc., SIMULINK Model-
interpolation to enhance computational inaccuracy based and system-based design, CD-ROM
gives good performances due to both power and Manual, Using Simulink Version 4, 2001.
three-phase fault disturbances. [11] The MathWorks Inc., Genetic Algorithms and
Direct Search TOOLBOX, CD-ROM Manual,
2004.
6 Acknowledgement [12] H. Sadaat, Power system analysis, McGraw-
The authors would like to acknowledge the financial Hill, 2004.
support from Suranaree University of Technology, [13] M. Rashidi, F. Rashidi, H. Monavar, Tuning of
Nakhon Ratchasima, THAILAND, during a period power system stabilizers via genetic algorithm
of this work. for stabilization of power system, IEEE Int.
Conf. Systems, Man and Cybernetics, Vol. 5, 5-
References: 8 October 2003, pp. 4649 – 4654.
[1] P. Kundur, Power system stability and control, [14] T. Kulworawanichpong, K-L. Areerak, K-N.
McGraw-Hill, 1994. Areerak, P. Pao-la-or, P. Puangdownreong, S.
[2] Y.H. Song, A.T. Johns, Flexible AC Sujitjorn, Dynamic parameter identification of
transmission systems (FACTS), The Institute of induction motors using intelligent search
Electrical Engineers, 1999. techniques, The 24th IASTED Int. Conf.
[3] A.H.M.A. Rahim, S.G.A. Nassimi, Synchronus Modelling, Identification, and Control (MIC
Generator Damping Enhancement Through 2005), Austria 2005, pp. 328-332.
Coordinated Control of Exciter and SVC, IEE [15] T. Charuwat, T. Kulworawanichpong, Genetic
Proceedings: Generation, Transmission and based distribution service restoration with
Distribution, Vol. 143, No. 2, 1996, pp. 211-218. minimum average energy not supplied, The 8th
[4] H.F. Wang, F.J. Swift, A Unified Model for the International Conference on Adaptive and
Analysis of FACTS Devices in Damping Power Natural Computing Algorithms
System Oscillations Part I: Single-machine (ICANNGA2007), 11–14 April 2007, pp. 230-
Infinite-bus Power Systems, IEEE Transaction 239.