Electrical Power and Energy Systems 29 (2007) 251259 www.elsevier.

com/locate/ijepes

A particle-swarm-based approach of power system stability enhancement with unied power ow controller
Ali T. Al-Awami a, Y.L. Abdel-Magid
a

b,*

, M.A. Abido

Electrical Engineering Department, King Fahd University of Petroleum & Minerals, Dhahran 31261, Box 784, Saudi Arabia b Electrical Engineering Program, The Petroleum Institute, P.O. Box 2533, Abu Dhabi, United Arab Emirates Received 29 July 2004; received in revised form 30 May 2006; accepted 19 July 2006

Abstract The use of the supplementary controllers of a unied power ow controller (UPFC) to damp low frequency oscillations in a weakly connected system is investigated. The potential of the UPFC supplementary controllers to enhance the dynamic stability is evaluated by measuring the electromechanical controllability through singular value decomposition (SVD) analysis. Individual designs of the UPFC controllers and power system stabilizer (PSS) using particle-swarm optimization (PSO) technique are discussed. The eectiveness of the proposed controllers on damping low frequency oscillations is tested through eigenvalue analysis and non-linear time simulation. 2006 Elsevier Ltd. All rights reserved.
Keywords: UPFC; Particle-swarm optimization; Power system stability

1. Introduction As power demand grows rapidly and expansion in transmission and generation is restricted with the limited availability of resources and the strict environmental constraints, power systems are today much more loaded than before. This causes the power systems to be operated near their stability limits. In addition, interconnection between remotely located power systems gives rise to low frequency oscillations in the range of 0.13.0 Hz. If not well damped, these oscillations may keep growing in magnitude until loss of synchronism results. Power system stabilizers (PSSs) have been used in the last few decades to serve the purpose of enhancing power system damping to low frequency oscillations. PSSs, which operate on the excitation system of generators, have proved to be ecient in performing their assigned tasks. However, PSSs may adversely aect voltage prole, may result in
Corresponding author. Tel.: +971 2 508 5506; fax: +971 2 508 5200. E-mail addresses: aliawami@kfupm.edu.sa (A.T. Al-Awami), yabdelmagid@pi.ac.ae (Y.L. Abdel-Magid). 0142-0615/$ - see front matter 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.ijepes.2006.07.006
*

leading power factor, and may not be able to suppress oscillations resulting from severe disturbances, especially those three-phase faults which may occur at the generator terminals. A wide spectrum of PSS tuning approaches has been proposed. These approaches have included pole placement [1], damping torque concepts [2], H1 [3], nonlinear and variable structure [4,5], and the dierent optimization and articial intelligence techniques [612]. FACTS devices have shown very promising results when used to improve power system steady-state performance. Through the modulation of bus voltage, phase shift between buses, and transmission line reactance, FACTS devices can cause a substantial increase in power transfer limits during steady-state. Because of the extremely fast control action associated with FACTS-device operations, they have been very promising candidates for utilization in power system damping enhancement. It has been observed that utilizing a feedback supplementary control, in addition to the FACTS-device primary control, can considerably improve system damping and can also improve system voltage prole, which is advantageous over PSSs.

252

A.T. Al-Awami et al. / Electrical Power and Energy Systems 29 (2007) 251259

A unied power ow controller (UPFC) is the most promising device in the FACTS concept. It has the ability to adjust the three control parameters, i.e. the bus voltage, transmission line reactance, and phase angle between two buses, either simultaneously or independently. A UPFC performs this through the control of the in-phase voltage, quadrature voltage, and shunt compensation. Till now, not much research has been devoted to the analysis and control of UPFCs. Several trials have been reported in the literature to model a UPFC for steady-state and transient studies. Under the assumption that the power system is symmetrical and operates under three-phase balanced conditions, Nabavi-Niaki and Iravani [13] developed a steady-state model, a small-signal linearized dynamic model, and a state-space large-signal model of a UPFC. Stefanov and Stankovic [14] developed an analytical large-signal model for unbalanced operation of the unied power ow controller (UPFC). Zhang et al. [15] developed a model for the Generalized Unied Power Flow Controller (GUPFC), which can control bus voltage and power ows of more than one line or even a sub-network using one shunt converter and two or more series converters. In 1999, Wang developed two UPFC models [16,17] which have been linearized and incorporated into the HeronPhillips model [18]. A number of control schemes has been suggested to perform the oscillation-damping task. Huang et al. [19] attempted to design a conventional xed-parameter leadlag controller for a UPFC installed in the tie line of a two-area system to damp the interarea mode of oscillation. Mok et al. [20] considered the design of an adaptive fuzzy logic controller for the same purpose. It has been illustrated through computer simulation that the proposed fuzzy logic controller outperforms the conventional xedparameter controller as it considers a wide range of operating conditions. Mishra et al. [21] and Schoder et al. [22] developed a TakagiSugeno (TS) type fuzzy logic controller for a UPFC to damp both local and interarea modes of oscillation for a multimachine power system. However, the initial parameters adjustment of this type of controller needs some trial and error. Dash et al. [23] suggested the use of a radial basis function NN for a UPFC to enhance system damping performance. The NN used either a single neuron or multiple neurons and the parameters were estimated using an error surface derived from the network inputs. Robust control schemes, such as H1 and singular value analysis, have also been explored. Vilathgamuwa et al. [24] proposed an H1 controller to regulate line currents in both the shunt and series UPFC inverters. The controller parameters have been selected based on the two-Riccati-equation approach. Pal [25] used the linearmatrix-inequality (LMI) formulation to approach the UPFC controller design based on H1 control scheme. To avoid pole-zero cancellation associated with the H1 approach, the structured singular value analysis have

been utilized in [26] to select the parameters of the UPFC controller to have the robust stability against model uncertainties. Recently, an integrated linear and nonlinear control of a UPFC for stability enhancement of a multimachine power system was developed [27]. Three pulse-width-modulation (PWM) UPFC parameters were controlled by the conventional PID structure, and the analytical expression of the nonlinear control law for the UPFC phase angle was obtained by a feedback linearization method. Furthermore, To avoid instability or loss of DC link capacitor voltage during transient conditions, Kannan et al. [28] proposed a real and reactive power coordination controller for a UPFC. In this paper, singular value decomposition (SVD) is used to select the control signal which is most suitable for damping the electromechanical (EM) mode oscillations. This is done as SVD analysis can be readily used to evaluate the EM mode controllability of the PSS and the dierent UPFC controllers. A SMIB system equipped with a PSS and a UPFC controller is used in this study. The problem of damping controllers design is formulated as an optimization problem to be solved using PSO. The aim of the optimization is to search for the optimum controller parameter settings that maximize the minimum damping ratio of the system. Eigenvalue analysis and non-linear simulation are used to assess the eectiveness of the proposed controllers to damp low frequency oscillations under dierent disturbances. 2. Problem statement Fig. 1 shows a SMIB system equipped with a UPFC. The UPFC consists of an excitation transformer (ET), a boosting transformer (BT), two three-phase GTO based voltage source converters (VSCs), and a DC link capacitors. The four input control signals to the UPFC are mE, mB, dE, and dB, where, mE is the excitation amplitude modulation ratio, mB is the boosting amplitude modulation ratio, dE is the excitation phase angle, and dB is the boosting phase angle.

v xtE

vEt iB
VSC-E VSC-B

vBt xB xBV BT

vb

it xE

iE

ET

vdc mE
E

mB

Fig. 1. SMIB power system equipped with UPFC.

A.T. Al-Awami et al. / Electrical Power and Energy Systems 29 (2007) 251259

253

2.1. Power system non-linear model By applying Parks transformation and neglecting the resistance and transients of the ET and BT transformers, the UPFC can be modeled as [13,16,17]: ! ! " mE cos dE vdc # ! vEtd 0 xE iEd 2 m sin 1 dE vdc E vEtq iEq xE 0 2 ! ! " mB cos dB vdc # ! vBtd 0 xB iBd 2 m sin 2 dB vdc B vBtq iBq xB 0 2 ! iEd 3mE _ dc 4Cdc cos dE sin dE v iEq 3 ! iBd 3mB 4Cdc cos dB sin dB iBq where vEt, iE, vBt, and iB are the excitation voltage, excitation current, boosting voltage, and boosting current, respectively; Cdc and vdc are the DC link capacitance and voltage, respectively. The non-linear model of the SMIB system of Fig. 1 is _ x b x 1 d _ P m P e Dx 1=M x _ 0 Ef d xd x0 id E0 =T 0 E
q d q

From (3), (8) and (9) we can nd iEd xBB 0 mE sin dE vdc xBd E xd R q 2xd R   xdE mB sin dB vdc vb cos d xd R 2 mE cos dE vdc xBq 2xqR   xqE mB cos dB vdc vb sin d xqR 2 xE 0 mE sin dE vdc xdE E xd R q 2x d R   xdt mB sin dB vdc vb cos d xd R 2 mE cos dE vdc xqE 2x  qR  xqt mB cos dB vdc vb sin d xqR 2

iEq

12

iBd

iBq

where xqR xq xtE xE xB xBV xE xq xtE xBq xB xBV xq xtE xqt xq xtE xE ; xqE xq xtE xd R x0d xtE xE xB xBV xE x0d xtE xBd xB xBV x0d xtE xdt x0d xtE xE ; xdE x0d xtE xBB xB xBV where xE, xB are the ET and BT reactances, respectively. 2.2. Power system linearized model The non-linear dynamic equations can be linearized around a given operating point to have the linear model given below: _ xb Dx Dd 13 _ DP m DP e DDx=M Dx 14 0 0 0 0 _ DE DEf d xd x Did =T 15 DE
q q d do

4 5 6 7
1=2 v2 ; vd q

do

_ f d K A V ref v uPSS Ef d =T A E v2 d

where Pe = vdid + vqiq, v xq iq , vq E0q 0 xd id , id = iEd + iBd, iq = iEq + iBq, where Pm and Pe are the input and output power, respectively; M and D the inertia constant and damping coecient, respectively; xb the synchronous speed; d and x the rotor angle and speed, respectively; E0q , E0f d , and v the generator internal, eld and terminal voltages, respectively; T 0d o the open circuit eld time constant; xd ; x0d , and xq the d-axis reactance, d-axis transient reactance, and q-axis reactance, respectively; KA and TA the exciter gain and time constant, respectively; Vref the reference voltage; and uPSS the PSS control signal. Also, from Fig. 1 we have  v jxtE vEt it   vEt  vBt jxBV  vb iB  8 9

_ f d DEf d K A DV ref Dv Dupss =T A DE _ dc K 7 Dd K 8 DE0q K 9 Dvdc Dv K ce DmE K cde DdE K cb DmB K cdb DdB where DP e K 1 Dd K 2 DE0q K pd Dvdc K pe DmE K pde DdE K pb DmB K pdb DdB DEq K 4 Dd K 3 DE0q K qd Dvdc K qe DmE K qde DdE K qb DmB K qdb DdB Dv K 5 Dd K 6 DE0q K vd Dvdc K ve DmE

16 17

where it and vb, are the armature current and innite bus voltage, respectively; vEt, vBt, and iB the ET voltage, BT voltage, and BT current, respectively. From (8) we have vd jvq xq iEq iBq jE0q x0d iEd iBd jxtE iEd iBd jiEq jiBq vEtd jvEtq where iE ET current. And from (9) we have vEtd jvEtq vBtd jvBtq jxBV iBd xBV iBq vb sin d jvb cos d 11 10

18 19

K vde DdE K vb DmB K vdb DdB 20 where K1K9, Kpu, Kqu, and Kvu are linearization constants. In state-space representation, the power system can be modeled as

254

A.T. Al-Awami et al. / Electrical Power and Energy Systems 29 (2007) 251259

_ Ax Bu x where the state vector x and control vector u are T x Dd Dx DE0q DEf d Dvdc u DuPSS DmE DdE Dmb Ddb 3 2 0 0 0 0 xb K pd 7 6 K D 2 M K 0 M 7 6 M1 M 7 6 K qd 7 6 K4 K3 1 0 T0 T0 7 A 6 T0 0 T do do do do 7 6 6 KAK5 K A K vd 7 1 AK6 0 KT 4 TA TA TA 5 A 2 K7 0 0
K pe M K T 0qe do T

21

In order to maintain the power balance between the series and shunt converters, a DC voltage regulator must be incorporated. The DC voltage is controlled through modulating the phase angle of the ET voltage, dE. Therefore, the dE damping controller to be considered is that shown in Fig. 4, where the DC voltage regulator is a PI-controller.

22

2.4. Objective function To select the best stabilizer parameters that enhance most the power system transient performance, the problem is formulated so as to optimize a selected objective function J subject to some inequality constraints, which are the maximum and minimum limits of each controller gain K and time constants T1T4. In this work, J minf Hence, the design problem can be formulated as maximize J Subject to K min 6 K 6 K max T min 6 T 1 6 T max 1 1 T min 6 T 2 6 T max 2 2 T min 6 T 3 6 T max 3 3 T min 6 T 4 6 T max 4 4 The proposed approach employs PSO to search for the optimum parameter settings of the given controllers. 24

K8 0
K p de M K T q0 de do

0 0
K pb M K T 0qb do

K 9 0
K p db M K Tq0 db do

6 60 6 6 B6 0 6 6 KA 4 TA 0

K ve KA TA

K vde KA TA

K vb KA TA

K ce

K cde

K cb

7 7 7 7 7 7 K A K v db 7 TA 5 K cdb

The linearized dynamic model of the state-space representation is shown in Fig. 2. In addition to the PSS input (uPSS), only one UPFC control input is considered in this Figure. 2.3. PSS and UPFC controllers The PSS structure to be considered is the very widely used lead-lag controller, whose transfer function is    sT w 1 sT 1 1 sT 3 uPSS K Dx 23 1 sT w 1 sT 2 1 sT 4 The UPFC damping controllers are of the structure shown in Fig. 3, where u can be mE, mB, or dB.

K1

1 D + sM s

K4 K2 E q
1 ' K 3 + sTdo

K5 E fd u pss

K6 -

KA 1 + sTA

K pu

K8

K qu u

K qd

K vu u

K vd Kpd

u K cu

vdc
9

s+K

K7
Fig. 2. Modied HeronPhillips linearized model.

A.T. Al-Awami et al. / Electrical Power and Energy Systems 29 (2007) 251259

255

Fig. 3. UPFC with lead-lag controller.

Fig. 4. UPFC with lead-lag controller and DC voltage regulator.

3. Controllability measure To measure the controllability of the EM mode by a given input (control signal), the singular value decomposition (SVD) is employed. Mathematically, if G is an m n complex matrix, then there exist unitary matrices W and V with dimensions of m m and n n, respectively, such that G W RV H where R R1 0 0 0 ! ; R1 diagr1 ; . . . ; rr

to control the mode associated with the eigenvalue k. Actually, the higher the rmin, the higher the controllability of this mode by the input considered. As such, the controllability of the EM mode can be examined with all inputs in order to identify the most eective one to control the mode. 4. Particle-swarm optimizer 4.1. Overview A novel population based optimization approach, called Particle-Swarm Optimization (PSO) approach, was introduced rst in [31]. This new approach features many advantages; it is simple, fast and can be coded in few lines. Also, its storage requirement is minimal. Moreover, this approach is advantageous over evolutionary and genetic algorithms in many ways. First, PSO has memory. That is, every particle remembers its best solution (local best) as well as the group best solution (global best). Another advantage of PSO is that the initial population of the PSO is maintained, and so there is no need for apply-

with r1 P P rr P 0 where r = min{m, n} and r1, . . ., rr are the singular values of G. The minimum singular value rr represents the distance of the matrix G from all the matrices with a rank of r 1. This property can be used to quantify modal controllability [29,30]. The matrix H can be written as H = [h1 h2 h3 h4 h5] where hi is a column vector corresponding to the ith input. The minimum singular value, rmin, of the matrix [kIAhi] indicates the capability of the ith input

256

A.T. Al-Awami et al. / Electrical Power and Energy Systems 29 (2007) 251259

ing operators to the population, a process that is time- and memory-storage-consuming. In addition, PSO is based on constructive cooperation between particles, in contrast with the genetic algorithms, which are based on the survival of the ttest [3134]. PSO starts with a population of random solutions particles in a D-dimension space. The ith particle is represented by Xi = (xi1, xi2, . . . , xiD). Each particle keeps track of its coordinates in hyperspace, which are associated with the ttest solution it has achieved so far. The value of the tness for particle i (pbest) is also stored as Pi = (pi1, pi2, . . . , piD). The global version of the PSO keeps track of the overall best value (gbest), and its location, obtained thus far by any particle in the population [31,32]. PSO consists of, at each step, changing the velocity of each particle toward its pbest and gbest according to Eq. (25). The velocity of particle i is represented as Vi = (vi1, vi2, . . . , viD). Acceleration is weighted by a random term, with separate random numbers being generated for acceleration toward pbest and gbest. The position of the ith particle is then updated according to Eq. (26) [31,32]. vid w vid c1 rand pid xid c2 Rand pgd xid 25 26

global minimum is better than gbest, then assign the current global minimum to gbest and assign the current coordinates (positions) to gbestx. Step. 7: Change the velocities according to Eq. (25). Step. 8: Move each particle to the new position according to Eq. (26) and return to Step 2. Step. 9: Repeat Step 28 until a stopping criterion is satised or the maximum number of iterations is reached. The proposed PSO-based approach was implemented using a MATLAB library built by the authors. In our implementation, the inertia weight, w, is linearly decreasing from 0.9 to 0.4, c1 and c2 are selected as 2, and the maximum number of iterations is 400. 6. Simulation results 6.1. Controllability measure SVD is employed to measure the controllability of the EM mode from each of the ve inputs: uPSS, mE, dE, mB, and dB. The minimum singular value, rmin, is estimated over a wide range of operating conditions. For SVD analysis, Pe ranges from 0.05 to 1.4 pu and Qe = [0.4, 0, 0.4]. At each loading condition, the system model is linearized, the EM mode is identied, and the SVD-based controllability measure is implemented. For comparison purposes, the minimum singular value for all inputs at Qe = 0.4, 0.0 and 0.4 pu is shown in Figs. 57, respectively. From these gures, the following can be noticed: EM mode controllability via dE is always higher than that of any other input. The capabilities of dE and mB to control the EM mode is higher than that of PSS.

xid xid vid

where, pid = pbest and pgd = gbest Several modications have been proposed in the literature to improve the PSO algorithm speed and convergence toward the global minimum. One modication is to introduce a local-oriented paradigm (lbest) with dierent neighborhoods [35]. It is concluded that gbest version performs best in terms of median number of iterations to converge. However, lbest version with neighborhoods of two is most resistant to local minima. PSO algorithm is further improved via using a time decreasing inertia weight, which leads to a reduction in the number of iterations [33]. 5. PSO algorithm Step. 1: Initialize an array of particles with random positions and their associated velocities to satisfy the inequality constraints. Step. 2: Check for the satisfaction of the equality constraints and modify the solution if required. Step. 3: Evaluate the tness function of each particle. Step. 4: Compare the current value of the tness function with the particles previous best value (pbest). If the current tness value is less, then assign the current tness value to pbest and assign the current coordinates (positions) to pbestx. Step. 5: Determine the current global minimum tness value among the current positions. Step. 6: Compare the current global minimum with the previous global minimum (gbest). If the current

Fig. 5. Minimum singular value with all stabilizers at Qe = 0.4.

A.T. Al-Awami et al. / Electrical Power and Energy Systems 29 (2007) 251259 Table 1 The optimal parameter settings of the individual controllers PSS K T1 T2 T3 T4 f 27.5 0.13 0.05 1.40 1.30 0.44 mE 12.75 0.56 0.05 1.06 0.76 0.15 dE 100 1.50 0.71 0.05 0.71 0.53 mB 36.99 1.29 0.67 0.15 0.05 0.44 dB

257

100 1.50 0.94 1.11 0.13 0.23

Table 2 Loading conditions Loading Nominal Heavy Pe (pu) 1.0 1.1 Qe (pu) 0.015 0.4

Fig. 6. Minimum singular value with all stabilizers at Qe = 0.0.

Also, notice that the optimization process has been carried out with the system operating at nominal loading condition given in Table 2. 6.3. Test results for individual designs To assess the eectiveness of the proposed stabilizers, two dierent loading conditions given in Table 2 were considered with dierent disturbances. It is worth mentioning that all the time domain simulation were carried out using the nonlinear power system model. The system data is given in the Appendix. 6.3.1. Nominal loading At this loading condition, the system eigenvalues with and without the proposed controllers are given in Table 3. It is clear that the open loop system is unstable. It is also clear that the proposed controllers stabilize the system. However, the value of f corresponding to the EM mode resulting from using PSS, dE or mB as control signals is much higher than those resulting from using dB or mE. This agrees with the SVD analysis carried out previously. The system behavior due to the utilization of the proposed controllers under transient conditions has been tested by applying a 6-cycle 3-phase fault at the innite bus at t = 1 s. The system response is shown in Fig. 8. The results conrm that mE and dB should not be used for enhancing system transient stability, a conclusion that can be readily drawn from SVD analysis. Hence, these two control signals are not considered for the coming tests. For comparison purposes, a GA-based PSS has been designed using the proposed technique. The resultant GA-based PSS parameters are as follows: K = 10.86, T1 = 0.92, T2 = 0.16, T3 = 0.44, T4 = 0.70, and the maximum damping ratio obtained using these parameters is f = 0.35. The performance of the GA-based PSS is compared with that of the PSO-based PSS whose parameters are shown in Table 1. Fig. 9 shows the system responses due to the use of the two stabilizers by applying a 6-cycle

Fig. 7. Minimum singular value with all stabilizers at Qe = 0.4.

The EM mode is more controllable with PSS than with either mE or dB. All control signals except mB at Qe = 0 and dE suer from low controllability to EM mode at low loading conditions.

6.2. Application of PSO to the design process The PSO algorithm has been applied to search for the optimal parameter settings of each of the supplementary controllers so that the objective function is optimized. It is worth mentioning that the DC voltage regulator gains, kdp and kdi, have been set a priori to the values shown in the Appendix. The nal parameter settings of the supplementary controllers and the corresponding maximum f are given in Table 1.

258

A.T. Al-Awami et al. / Electrical Power and Energy Systems 29 (2007) 251259

Table 3 System eigenvalues with and without control (nominal loading conditions) No control 1.5 j5.3, 0.2723 11.5 j6.9 15.5, 5.1 PSS 3.6 j7.3, 0.4423 3.5 j7.1 26.3, 15 5.0, 0.8 0.2068 mE 1.1 j6.9, 0.1574 1.2 j7.4 28 j6.2 14.7, 5.2 1.3, .2 dE 2.6 j4.2, 0.5264 4.1 j5.8 15.5 8.4, 5.1 1.1, .2 mB 3.2 j6.5, 0.4417 3.1 j6.3 24 j11.7 15.5, .2 5.1, 1.8 dB 1.5 j6.4, 0.2282 1.4 j6.0 21 j8.2 15.3, .2 5.1, 1.1

Fig. 8. System response to 6-cycle fault disturbance for nominal loading. Fig. 10. System response to 3-cycle fault disturbance for heavy loading.

From the above conducted tests, it can be concluded that the dE controller is the most robust controller, which agrees with the SVD analysis. 7. Conclusion In this paper, SVD has been employed to evaluate the EM mode controllability to PSS and the four UPFC control signals. It has been shown that the EM mode is most strongly controlled via dE for a wide range of loading conditions. In addition, SVD analysis has illustrated that the EM mode is poorly controlled through mE and dB. An optimization technique has been proposed to design the PSS and UPFC controllers individually. PSO has been utilized to search for the optimal controller parameter settings that optimize an eigenvalue-based objective function. Simulation results through nonlinear power system model has proved the conclusions drawn from SVD analysis. Acknowledgement 6.3.2. Heavy loading A 3-cycle 3-phase fault has been applied at the innite bus at this loading condition. The system response is shown in Fig. 10. The authors would like to acknowledge King Fahd University of Petroleum and Minerals for the support of this work.

Fig. 9. PSO- vs. GA-based PSS performances to 6-cycle fault disturbance for nominal loading.

3-phase fault at the innite bus at t = 1 s. The test results demonstrate clearly that the proposed PSO-based stabilizer outperforms the GA-based stabilizer.

A.T. Al-Awami et al. / Electrical Power and Energy Systems 29 (2007) 251259

259

Appendix. The test system parameters are: Machine: xd = 1; xq = 0.6; x0d 0:3; D = 0; M = 8.0; T 0d o 5:044; freq = 60; v = 1.05; Exciter: KA = 50; TA = 0.05; Efd_max = 7.3; Efd_min = 7.3; PSS: Tw = 5; Ti_min = 0.05; Ti_max = 1.5; i = 1, 2, 3, 4; Kmin = 0.0; Kmax = 100.0; uPSS_max = 0.2; uPSS_min = 0.2; DC voltage regulator: kdp = 10; kdi = 0; Transmission Line: xtE = 0.1; xBV = 0.6; UPFC: xE = 0.1; xB = 0.1; Ks = 1; Ts = 0.05; Cdc = 3; Vdc = 2; mE_max = 2; mE_min = 0; mB_max = 2; mB_min = 0. References
[1] Chen CL, Hsu YY. Coordinated synthesis of multimachine power system stabilizer using an ecient decentralized modal control (DMC) algorithm. IEEE Trans Power Sys 1987;9(3):54351. [2] Gibbard MJ. Co-ordinated design of multimachine power system stabilisers based on damping torque concepts. IEE Proc Pt C 1988;135(4):27684. [3] Klein M, Le LX, Rogers GJ, Farrokhpay S, Balu NJ. H1 damping controller design in large power systems. IEEE Trans Power Sys 1995;10(1):15866. [4] Lu Q, Sun YZ. Nonlinear stabilizing control of multimachine systems. IEEE Trans Power Sys 1989;4(1):23641. [5] Samarasinghe VGDC, Pahalawaththa NC. Damping of multimodal oscillations in power systems using variable structure control techniques. IEE Proc Gen Trans Distrib 1997;144(3):32331. [6] Taranto JN, do Bomm ALB, Falcao DM, Martins N. Automated design of multiple damping controllers using genetic algorithms. IEEE Power Eng Society 1999 Winter Meeting, vol. 1, 31 January4 February 1999, p. 53944. [7] Abdel-Magid YL, Abido MA, Al-Baiyat S, Mantawy AH. Simultaneous stabilization of multimachine power systems via genetic algorithms. IEEE Trans Power Sys 1999;14(4):142839. [8] Abdel-Magid YL, Abido MA, Mantawy AH. Robust tuning of power system stabilizers in multimachine power systems. IEEE Trans Power Sys 2000;15(2):73540. [9] Abido MA. Particle swarm optimization for multimachine power system stabilizer design. IEEE Power Eng Society Summer Meeting, vol. 3, 1519 July 2001, p. 134651. [10] Abido MA, Abdel-Magid YL. Robust design of electrical powerbased stabilizers using tabu search. IEEE Power Eng Society Summer Meeting, vol. 3, 1519 July 2001, p. 157378. [11] Abido MA, Abdel-Magid YL. Optimal design of power system stabilizers using evolutionary programming. IEEE Trans Energ Conver 2002;17(4):42936. [12] Abido MA, Abdel-Magid YL. Radial basis function network based power system stabilizers for multimachine power systems. Int Conf Neural Networks 1997;2(912):6226. [13] Nabavi-Niaki A, Iravani MR. Steady-state and dynamic models of unied power ow controller (UPFC) for power system studies. IEEE Trans Power Sys 1996;11(4):193743. [14] Stefanov PC, Stankovic AM. Modeling of UPFC operation under unbalanced conditions with dynamic phasors. IEEE Trans Power Sys 2002;17(2):395403.

[15] Zhang XP, Handschin E, Yao M. Modeling of the generalized unied power ow controller (GUPFC) in a nonlinear interior point OPF. IEEE Trans Power Sys 2001;16(3):36773. [16] Wang HF. Damping function of unied power ow controller. IEE Proc Gen Trans Distrib 1999;146(1):817. [17] Wang HF. Application of modeling UPFC into multi-machine power systems. IEE Proc Gen Trans Distrib 1999;146(3):30612. [18] Heron WG, Phillips RA. Eect of modern amplidyne voltage regulators on under-excited operation of large turbine generator. IEEE Trans Power Apparatus Sys 1952:6927. [19] Huang Zhengyu, Ni Yinxin, Shen CM, Wu FF, Chen Shousun, Zhang Baolin. Application of unied power ow controller in interconnected power systems-modeling, interface, control strategy, and case study. IEEE Trans Power Sys 2000;15(2):81724. [20] Mok TK, Ni Y, Wu FF. Design of fuzzy damping controller of UPFC through genetic algorithm. IEEE Power Eng. Society Summer Meeting, vol 3, 1620 July 2000, p. 188994. [21] Mishra S, Dash PK, Panda G. TS-fuzzy controller for UPFC in a multi-machine system. IEE Proc Gen Trans Distrib 2000;147(1): 1522. [22] Schoder K, Hasanovic A, Feliachi, A. Power system damping using fuzzy controlled unied power ow controller, IEEE PES Winter Meeting, 2001. vol. 2, 28 January1 February 2001, p. 61722. [23] Dash PK, Mishra S, Panda G. A radial basis function neural network controller for UPFC. IEEE Trans Power Sys 2000;15(4): 12939. [24] Vilathgamuwa M, Zhu X, Choi SS. A robust control method to improve the performance of a unied power ow controller. Electr Power Sys Res 2000;55:10311. [25] Pal BC. Robust damping of interarea oscillations with unied power ow controller. IEE Proc Gen Trans Distrib 2002;149(6):7338. [26] Seo Jang-Cheol, Moon Seung-Il, Park Jong-Keun, Choe JongWoong. Design of a robust UPFC controller for enhancing the small signal stability in the multi-machine power systems. IEEE PES Winter Meeting, vol. 3, 28 January1February 2001, p. 1197202. [27] Xie H, Xu Z, Lu Q, Song YH, Yokoyama A, Goto M. Integrated linear and nonlinear control of unied power ow controllers for enhancing power system stability. Electr Power Compo Sys 2003;31:33547. [28] Kannan S, Jayaram S, Salama MMA. Real and reactive power coordination for a unied power ow controller. IEEE Trans Power Sys 2004;19(3):145461. [29] Hamdan AMA. An investigation of the signicance of singular value decomposition in power system dynamics. Int J Electr Power Energ Sys 1999;21:41724. [30] Abdel-Magid YL, Abido MA. Robust coordinated design of excitation and TCSC-based stabilizers using genetic algorithm. Electr Power Sys Res 2004;69:12941. [31] Kennedy J, Eberhart R. Particle swarm optimization. Proc IEEE Int Conf Neural Networks 1995;4:19428. [32] Eberhart R, Kennedy J. A new optimizer using particle swarm theory, In: Proc Sixth Int Symp Micro Machine Human Science, vol. 46 October, 1995, p. 3943. [33] Shi Y, Eberhart R. A modied particle swarm optimizer. In: IEEE Int Conf Evol Computation Proc, IEEE World Congress Computational Intelligence., 49 May, 1998, p. 6973. [34] Shi Y, Eberhart RC. Empirical study of particle swarm optimization, In: Proc 1999 Congress Evol Computation, vol. 3 , 69 July, 1999, p. 1950. [35] Suganthan PN. Particle swarm optimiser with neighbourhood operator. In: Proc 1999 Congress Evol Computation, vol. 3, 69 July, 1999, p. 1962.