Professional Documents
Culture Documents
Abstract
Optimal multiobjective design of robust multimachine power system stabilizers (PSSs) using
simulated annealing optimization technique is presented in this paper. The proposed approach
employs Simulated Annealing (SA) optimization technique to search for optimal parameter settings
of a widely used conventional fixed-structure lead-lag PSS (CPSS). A multiobjective problem is
formulated to optimize a composite set of objective functions comprising the damping factor, and the
damping ratio of the lightly damped electromechanical modes. The effectiveness of the proposed
technique in damping local and inter area modes of oscillations in multimachine power systems,
over a wide range of loading conditions and system configurations, is confirmed through eigen
value analysis and nonlinear simulation results.
July Issue
Page 37 of 92
2. Statement of Problem
A. Power System model
The generator in the power system is represented by fourth order model and the problem is to design
the parameters of the power system stabilizers so as to stabilize a system of N generators
simultaneously. The fourth order power system model is represented by a set of non-linear differential
equations given for any ith machine,
d i
dt
di
dt
i s
s
2H
(1)
( Pmi Pei )
(2)
'
dEqi
1
' I (X X ' ) E ]
[ Eqi
di di
di
fdi
dt
Td0
i
'
dEdi
1
' I ( X X ' )]
[ Edi
qi qi
qi
'
dt
Tq0i
dE fdi
1
[ E fdi Kai (Vrefi Vti )]
Tai
' I ( x' x ' ) I I
Tei E'di I di Eqi
qi
qi di di qi
dt
where
(3)
(4)
(5)
(6)
E fdi , I di and I qi are the field voltage, d-axis stator current and q- axis stator current,
July Issue
Page 38 of 92
i ( s)
1 sTwi (1 sT2i ) (1 sT4i )
(7)
Where is the deviation of the speed of the rotor from synchronous speed
The second term in Eq. (6) is the washout term with a time constant of Tw . The third term is the lead
lag compensation to counter the phase lag through the system. The washout block serves as a high-pass
filter to allow signals in the range of 0.22.0 Hz associated with rotor oscillations to pass unchanged.
This can be achieved by choosing a high value of time constant ( Tw ). However, it should not be so
high that, it may create undesirable generator voltage excursions during system-islanding [12].
Compromising, it may have a value anywhere in the range of 120 s [12]. On the other hand, the lead
lag block present in the system provides phase lead (some rare cases lag also) compensation for the
phase lag that is introduced in the circuit between the exciter input (i.e. PSS output) and the electrical
torque. In this study the parameters to be optimized are{ K i , T1i , T2i ; i=1,2 3,...m },assuming
j 1 i, j 0
(8)
Where np is the number of operating points considered in the design process, and i, j is the real
part of the ith eigen value of the jth operating point, subject to the constraints that finite bounds are
placed on the power system stabilizer parameters. The relative stability is determined by the value of
0 . This will place the closed-loop eigen values in a sector in which as shown in Fig. 1.
j
i, j 0
2) To limit the maximum overshoot, the parameters of the PSS may be selected to minimize the
following objective function:
July Issue
Page 39 of 92
j 1 i, j 0
(9)
where i, j is the damping ratio of the i eigen value of the j operating point. This will place the
th
th
3) The single objective problems described may be converted to a multiple objective problem by
assigning distinct weights to each objective. In this case, the conditions i, j 0 and i, j 0 are
imposed simultaneously. The parameters of the PSS may be selected to minimize the following
objective function:
J J1 a. J 2
np
np
=
[ 0 i, j ]2 + a
[ 0 i, j ]2
j 1 i, j 0
j 1 i, j 0
(9)
This will place the system closed-loop eigen values in the D-shape sector characterized by i, j 0
and i, j 0 as shown in Fig. 3.
j
i, j 0
i, j 0
It is necessary to mention here that only the unstable or lightly damped electromechanical modes of
oscillations are relocated. The design problem can be formulated as the following constrained
optimization problem, where the constraints are the PSS parameter bounds:
Minimize J subject to
Ki
Ki Ki
min
max
T1i
T1i T1i
min
max
T2i
T2i T2i
min
max
July Issue
Page 40 of 92
3. Simulated Annealing
The Simulated annealing method resembles the cooling process of molten metals through annealing. At
high temperature, the atoms in the molten metal can move freely with respect to each other, but as the
temperature is reduced, the movement of the atoms gets restricted. The atoms start to get ordered and
finally form crystals having the minimum possible energy. However the formation of the crystal mostly
depends on the cooling rate. If the temperature is reduced at a fast rate, the crystalline state may not be
achieved at all, instead, the system may end up in a polycrystalline state, which may have a higher
energy state than the crystalline state. Therefore, in order to achieve the absolute minimum energy
state, the temperature needs to be reduced at a slow rate. The process of slow cooling is known as
annealing in metallurgical parlance. The simulated annealing procedure simulates this process of slow
cooling of molten metal to achieve the minimum function value in a minimization problem. The
cooling phenomenon is simulated by controlling a temperature-like parameter introduced with the
concept of Boltzmann probability distribution. According to the Boltzmann probability distribution, a
system in thermal equilibrium at a temperature T has its energy distributed probabilistically according
to Eqn. (10)
-E
P(E ) exp
(10)
kT
where k is the Boltzmann constant. It can be seen from Eqn.(10) that system at a high temperature has
almost uniform probability of being at any energy state but at low temperature it has a small probability
of being at a high energy state. Therefore by controlling the temperature T and assuming that the
search process follows the Boltzmann probability distribution, the convergence of the algorithm can be
controlled. If at any instant the current point is xt and the function value at that point is E(t ) f ( xt ) ,
then the probability of the next point being at xt 1 depends on the difference in the function values at
these two points i.e. E E(t 1) E(t ) and is calculated using the Boltzmann probability distribution
given by Eqn.(11)
- E
P( E (t 1)) min1, exp
(11)
kT
If E 0 , this probability is one and the point xt 1 is accepted. But if E 0 it implies that
xt 1 is worse than that of xt . In this case according to Metropolis algorithm, there is some finite
probability of selecting the point xt 1 even though it is worse than xt . This probability depends on
relative magnitude of E and T . If the parameter T is large, this probability is high for points with
large function values. Thus any point is acceptable for large values of T . On the other hand, if the
parameter T is small, the probability of accepting an arbitrary point is small. Thus for small values of
T , the points with only small deviation in function value are accepted.
July Issue
Page 41 of 92
G2
2
5
Load A
3
6
Load B
4
1
G1
Figure 4. WSCC Three-machine Nine-bus Power System
Load
A
B
C
Gen
1
2
3
Case 3
P
Q
1.50
0.90
1.20
0.80
1.00
0.50
Case 3
P
0.33
2.0
1.50
Q
1.12
0.57
0.38
4.3321
2.463
0.3997
0.4057
0.3716
0.3752
0.2739
0.2990
0.2961
3.40
7.00
1.60
0.3067
0.3067
0.3533
0.1133
0.1667
0.2600
Table 5: Comparison of Eigen values and Damping ratios for different cases
Base Case
Case-1
Case-2
Case-3
July Issue
Without PSS
-0.2367 8.5507i, 0.0277
-11.1752 10.4687i, 0.7298
-0.1421 8.4615i, 0.0168
-11.2788 11.3006i, 0.7064
-0.8199 8.1535i, 0.1001
-10.4600 12.2400i, 0.6497
0.0990 + 8.5483i, -0.0116
-11.4841 +11.0256i, 0.7214
CPSS
-0.8017 9.0603i, 0.0881
-11.1414 9.4032i, 0.7642
-0.8024 8.9184i, 0.0896
-11.1601 10.3813i, 0.7322
-1.2583 8.4817i, 0.1468
-10.3426 11.4081i, 0.6717
-0.3549 + 8.9847i, 0.0395
-11.3684 +10.0945i, 0.7478
Page 42 of 92
SA PSS
-4.0591 5.7725i, 0.5752
-7.9563 9.2689i, 0.6513
-3.3263 6.0414i, 0.4823
-8.6204 9.6271i, 0.6671
-4.2574 8.4693i, 0.4491
-5.9496 7.1843i, 0.6378
-3.8036 + 6.3262i, 0.5153
-8.4053 +10.0518i, 0.6415
Figure 5. Speed deviation of 2nd and 3rd generators for Case (a)
Case(b): A 6-cycle fault disturbance at bus 7 at the end of line 57with case 3. The fault is cleared by
tripping the line 57 with successful reclosure after 1.0 s
Figure 6. Speed deviation of 2nd and 3rd generators for Case (b)
The system responses to the considered faults with and without the proposed SAPSSs are shown in
Figs. 5 and 6 respectively. It is clear that the proposed SAPSSs provide good damping characteristics
to low frequency oscillations and greatly enhance the dynamic stability of power systems.
Test System-II
To demonstrate the effectiveness of the proposed method on a larger and more complicated power
system, the readily accessible 10-generator 39-bus New England system is adopted. Fig. 7 shows the
configuration of the test system. All generating units are represented by fourth-order model and their
static exciters are equipped with PSS. Details of the system data are given in [17].
July Issue
Page 43 of 92
The tuned parameters of the ten PSS using conventional root locus approach and proposed simulated
annealing optimization method are shown in the Table 6.
The small signal analysis of the test system was carried out without connecting the PSS. The
electromechanical modes and the damping ratios obtained for all the above cases with the proposed
approach and CPSS in the system are given in Table 7. The unstable modes for different operating
conditions were found out and highlighted in the above Table.
8
37
26
30
28
25
29
27
38
24
18
17
16
6
21
15
3
4
10
35
22
14
39
12
19
5
6
13
11
31
23
10
2
8
20
33
32
3
34
36
7
July Issue
Page 44 of 92
CPSS Parameters
K
10.4818
0.6799
0.2396
1.1531
17.0819
13.4726
4.3773
0.5709
1.6059
19.8488
Base Case
(All lines
in service)
T1
0.6211
0.6185
0.5778
0.5727
0.6143
0.6163
0.5636
0.6099
0.5429
0.5027
SA PSS Parameters
T2
0.1789
0.1796
0.1923
0.1940
0.1809
0.1803
0.1971
0.1822
0.2046
0.2210
K
8.4667
14.0667
14.0667
12.2000
10.3333
6.6000
6.6000
2.8667
8.4667
8.4667
T1
0.5267
0.5267
0.6067
0.6600
0.5533
0.7133
0.8733
0.6600
0.6333
0.6867
T2
0.1133
0.1133
0.2333
0.2067
0.2067
0.2067
0.1400
0.2200
0.2200
0.3000
Table 7. Comparison of eigen values and damping ratios for different cases
Without PSS
CPSS
SA PSS
-1.1878 +10.6655i, 0.1107
-1.5226 +11.7232i, 0.1288
-4.1286 +12.4994i, 0.3136
-0.3646 + 8.8216i, 0.0413
-1.3326 +11.2726i, 0.1174
-1.4869 +11.4295i, 0.1290
-0.3063 + 8.5938i, 0.0356
-1.9859 +11.1499i, 0.1753
-0.9327 + 9.4601i, 0.0981
-0.2718 + 8.1709i, 0.0332
-0.9837 + 9.0350i, 0.1082
-1.1302 + 9.6094i, 0.1168
-0.0625 + 7.2968i, 0.0086
-0.5380 + 8.5014i, 0.0632
-1.2623 + 9.1152i, 0.1372
-0.1060 + 6.8725i, 0.0154
-0.1568 + 7.3758i, 0.0213
-1.3386 + 8.0520i, 0.1640
-1.0658 + 7.2601i, 0.1452
-3.2689 + 7.3285i, 0.4074
0.2579 + 6.1069i, -0.0422
-2.9298 + 4.1507i, 0.5767
0.0620 + 6.1767i, -0.0100
-0.0046 + 6.3800i, 0.0007
-1.2016 + 4.5676i, 0.2544
-1.1338 + 3.4802i, 0.3098
0.0794 + 3.9665i, -0.0200
Case-1
-1.5173 +11.7109i,
-1.3362 +11.2695i,
-1.9880 +11.1547i,
-0.9669 + 9.0331i,
-0.1593 + 7.3687i,
-0.0826 + 6.1146i,
-1.0081 + 6.0958i,
-1.9766 + 6.0065i,
-1.3318 + 3.1517i,
0.1285
0.1177
0.1755
0.1064
0.0216
0.0135
0.1632
0.3126
0.3892
-4.1613 +12.4079i,
-1.4918 +11.4276i,
-0.9094 + 9.4835i,
-1.3580 + 9.2344i,
-1.2435 + 9.1003i,
-1.3980 + 8.0167i,
-3.3198 + 7.3557i,
-2.3455 + 3.3520i,
-1.2008 + 3.3464i,
Case-2
-1.3328 +11.2585i,
-1.4566 +11.2977i,
-2.0073 +11.0646i,
-0.5655 + 8.4628i,
-0.8322 + 8.1841i,
-0.1551 + 7.3961i,
-0.0043 + 6.3664i,
-2.2257 + 6.2886i,
-1.3470 + 5.5976i,
0.1176
0.1279
0.1785
0.0667
0.1012
0.0210
0.0007
0.3336
0.2340
-1.3405 +11.3267i,
-1.3380 +11.2101i,
-2.0206 +11.0315i,
-0.5650 + 8.4482i,
-0.7508 + 8.1182i,
-0.1506 + 7.4154i,
-0.0023 + 6.3596i,
-0.6910 + 5.8629i,
-0.7668 + 3.3898i,
0.1175
0.1185
0.1802
0.0667
0.0921
0.0203
0.0004
0.1171
0.2206
-4.1269 +12.5528i,
-1.4705 +11.3578i,
-0.9902 + 9.4536i,
-1.1759 + 9.3214i,
-3.4864 + 7.4509i,
-1.3664 + 8.0023i,
-1.4953 + 7.7485i,
-2.5309 + 4.2218i,
-0.9499 + 3.3941i,
Case-3
July Issue
Page 45 of 92
0.3180
0.1294
0.0955
0.1455
0.1354
0.1718
0.4114
0.5733
0.3377
0.3123
0.1284
0.1042
0.1252
0.4238
0.1683
0.1895
0.5142
0.2695
and
are chosen to be -1.0 and 0.2 respectively. Several values for weight a
Figure 8. Speed deviations of 5th and 6th generators for Case (a)
Figure 9. Speed deviations of 2nd and 3rd generators for Case (b)
July Issue
Page 46 of 92
Figure 10. Speed deviations of 8th and 9th generators for Case (c)
In all the above cases, the system performance with the proposed SAPSSs is much better than that of
CPSSs and the oscillations are damped out much faster. In addition, the proposed SAPSSs are quite
efficient to damp out local and inter area modes of oscillations. This illustrates the potential and
superiority of the proposed design approach to get optimal set of PSS parameters.
5. Conclusions
In this study, optimal multiobjective design of robust multi-machine power system stabilizers (PSSs)
using SA is proposed. The approach effectiveness is validated on two multi-machine power systems. In
this paper, the performance of proposed SAPSS is compared with conventional speed-based lead-lag
PSS. A multiobjective problem is formulated to optimize a composite set of objective functions
comprising the damping factor, and the damping ratio of the lightly damped electromechanical modes.
The problem of robustly selecting the parameters of the power system stabilizers is converted to an
optimization problem which is solved by SA with the eigen value-based multiobjective function.
Eigen value analysis under different operating conditions reveal that undamped and lightly damped
oscillation modes are shifted to a specific stable zone in the s-plane. These results show the potential of
SA algorithm for optimal settings of PSS parameters.
The nonlinear time-domain simulation results show that the proposed PSSs work effectively over a
wide range of loading conditions and system configurations.
6. References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
July Issue
Page 47 of 92
Authors Profile
G.Naresh graduated from Andhra University in 2001, Masters in 2004 from JNT
University and currently pursuing Ph.D. from JNT University Kakinada, INDIA.
Presently he is an Associate Professor in the Department of Electrical & Electronics
Engineering, Pragati Engineering College, Surampalem, AP since June 2003. His
research interests include Power System Stability, Power System Operation &
Control and Applications of Evolutionary Computing Techniques to Electrical
Engineering.
Your photo
Comes
here
Your photo
Comes
here
Dr. M.Ramalinga Raju graduated in 1986 from JNTU, Masters in 1989 from REC,
Warangal and Ph.D. in 2004 from JNT University, INDIA. Presently he is Professor
and Head of Electrical & Electronics Engineering Department, University College of
Engineering Kakinada, JNTUK. He presented many research papers in various
national and international conferences and journals. His areas of interests include
Energy Management, Conservation and Auditing, Distributed Generation and IT
Applications in Power Utility Companies.
Dr. S.V.L. Narasimham graduated in 1982, Masters in 1987, 1995 and Ph.D. in
2000, all from JNT University, INDIA. Presently he is working as Professor of
Computer Science and Engineering, School of Information Technology, JNTUH,
INDIA. He presented more than 60 research papers in various national and
international conferences and Journals. His areas of interest include Energy
Optimization and Audit, Real time Power System Operation and Control, Image
Processing, Character Recognition, Home Automation and e- Governance.
Your photo
Comes
here
K.Ravindra graduated from Nagarjuna University in 2001, Masters from PSG
College of Engineering & Technology, Bharathiar University in 2003. Presently he
is an Assistant Professor in Electrical & Electronics Department, University College
of Engineering, JNTU Kakinada, since 2003. His research interests include
Distributed Generation and Distribution system planning.
Your photo
July IssueComes
here
Page 48 of 92