You are on page 1of 7

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/251987294

Analysis and design of multi-stage LQR UPFC

Article · November 2010


DOI: 10.1109/ICPCES.2010.5698644

CITATIONS READS
12 159

1 author:

Rajendra Kumar Pandey


Indian Institute of Technology (Banaras Hindu University) Varanasi
45 PUBLICATIONS   233 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Smart Power Control View project

All content following this page was uploaded by Rajendra Kumar Pandey on 06 June 2016.

The user has requested enhancement of the downloaded file.


1

Analysis and Design of Multi-Stage LQR


UPFC
R. K. Pandey, Senior Member IEEE

These oscillations are due to the dynamics of inter-area power


Abstract-- This paper presents stability analysis and design of transfer and often exhibit poor damping when aggregate
multi-stage LQR unified power flow controller (UPFC) for multi- power transfer over a transmission corridor is high relative to
machine power system based on generalized modelling approach. the transmission strength. Controllers must operate
An improved LQR control has been proposed as Multi-Stage satisfactorily in the presence of many modes of power swings
LQR which considers the predominant eigenvalues of system for and also over a wide range of operation. So, there is need of
selection of weights of Q matrix that gives very fast tracking for a
power system oscillations damping (POD) device for
given disturbance. The approach is modular and general enough
so as to include any sub system along with detailed oscillation damping so as to maintain system stability along
representation. The performance of multi-stage LQR UPFC has with reliability of system operations. Initially, conventional
been evaluated for a sample two area system which can readily power system stabilizer (PSS) was used to damp out the local
be extended to multi-area with some augmentation. mode of oscillations but due to increasing demand of power
supply and contingency problem, PSS functioning has
Index Terms─ Damping controllers, flexible AC transmission deteriorated. Since, the network power balance has to be
systems (FACTS), multi-stage LQR, Phillips-Heffron model, achieved by modulating power flow to ensure smooth function
power system stabilizers (PSS), static synchronous series of existing controllers; this can only be ensured by placing
compensator (SSSC), static synchronous shunt compensator
adequate controller in network which will enhance power
(STATCOM), unified power flow controller (UPFC).
transfer capacity of the transmission line within acceptable
time, thus maintaining stable system operation.
I. INTRODUCTION Modern control components such as Flexible AC
Transmission System (FACTS) devices, not only helps in
eregulation, environmental legislations and cost of
D
increasing operational stability but also adds to enhanced
construction has constrained building of new network reliability. UPFC has been proved beneficial to
transmission lines. It is desirable to load existing provide adequate control of important parameters of power
transmission lines close to their thermal limits. For network [1-3]. UPFC may be a versatile and emerge as a
economical and technological reasons, individual power strong candidate for power control. It combines features of
systems are organized in the form of electrically connected Static Synchronous Compensator (STATCOM) and Static
areas or regional grids. They are eventually interconnected Synchronous Series Compensator (SSSC) [5, 6]. UPFC is
and contractually tied with certain generation and multifunctional FACTS device with capability of controlling
scheduling features. In present scenario of growing power all three parameters that dictate power flow over power
system network, stability has become a thought of primary transmission line, i.e., line voltage, impedance and phase
concern, as power oscillations are very common phenomenon angle, sequentially or concurrently, with internal reactive
due to different operating and changing conditions prevailing power generation in real time, it can be used for effective and
in the network. An improperly designed controller would not efficient power flow control, enhancement of transient
only degrade the system performance but may also aggravate stability, mitigation of low frequency power system
adverse dynamic interactions [17]. Therefore, proper control oscillations and voltage (reactive power) regulation. Different
design with adequate strategies should be adopted for approaches based on modern control theory have been applied
damping oscillations to ensure stability. Also, in order to in POD controller design. Wang et. al. have presented a
operate power system effectively, without reduction in the modified linearized Phillips-Heffron model of a power system
system security and quality of supply, even in the case of installed with UPFC and addressed basic issues pertaining to
contingency conditions, new control strategies need to be design of UPFC based power oscillation damping controller
implemented. along with selection of input parameters of UPFC to be
A problem of interest, at present has been in power modulated in order to achieve desired damping [7]. T.K. Mok
industry to mitigate low frequency oscillations which often et. al. have presented fuzzy damping controller of UPFC
arise among areas in a large interconnected power network. through genetic algorithm [8]. L.Y. Dong et. al. have proposed
a PI based approach for the dynamic control of UPFC with
R. K. Pandey is with Electrical Engineering Dept, Institute of Technology,
new control strategy, the active and reactive power flow
Banaras Hindu University, Varanasi, India. Email: rpsneh@yahoo.co.in control was achieved ensuring damping of system oscillations

978-1-4244-8542-0/10/$26.00 ©2010 IEEE


[9]. Z. Huang et. al. have tested two damping schemes for PWM technique has been considered. Where input control
UPFC, one is voltage modulation in voltage control of shunt signals me , mb and δ e , δ b known as amplitude modulation
element and other is power modulation in constant power ratio and phase angle of each VSC respectively.
control of series element, a cascade lead-lag transfer function On neglecting the converter harmonics, the following
has been used for the supplementary control [10]. equations can be written relating the voltage at AC side
However, FACTS controller design in power transmission terminal of shunt and series connected converters VSC-E and
network requires a state space model in a generalized VSC-B to the capacitor voltage Vdc on the common dc-link.
framework which may be used to assess the capabilities of
individual controller in changing system operating conditions. _ ⎫
This can be used as analysis tool. This paper presents design V E = V Ed + jV Eq = kmeV dc e jδ e ⎪⎪
and analysis of a novel multi-stage LQR UPFC utilizing state ⎬ (1)
_
space model developed. jδ b ⎪
V B = V Bd + jV Bq = kmbV dc e ⎭⎪
II. UPFC STATE SPACE MODEL where, k = ratio between AC and DC voltage depending on
This section presents model development in state space structure of converter. The instantaneous power at the AC and
framework. The system comprising of multi machines can be DC terminals of shunt and series connected converters VSC-E
conceptualized as shown in Fig 1 and VSC-B are equal, if converters are assumed to be lossless.
Hence power balance equation is
V dc .I i = V Ed .I Ed + V Eq .I Eq ⎫⎪
⎬ (2)
V dc .I o = V Bd .I Bd + V Bq .I Bq ⎪⎭

Since the net current to capacitor is zero, the DC circuit can be


described as
dVdc
Cdc = ( Ii + Io ) (3)
dt
where Ii and Io are current at VSC-E and VSC-B end
Fig. 1. Multi-machine representation in power network respectively.
Multiple Generators of respective end may be grouped as
area and thus a two-area power system with UPFC can be
represented as shown in Fig.2 and Fig 3.

Fig. 3. Two-area power network with UPFC


For Fig. 1 the following equations are written

dVdc kme ⎡I + I ⎤
= [cosδ e sin δ e ].⎢ I1d + I 2d ⎥
dt Cdc ⎣ 1q 2q ⎦
(4)
km ⎡− I 2d ⎤
Fig. 2. UPFC installed in a two area system + b [cosδ b sin δ b ].⎢ ⎥
Cdc ⎣ − I 2q ⎦
Initially, an UPFC installed in two-area power network
connected via an intertie has been modelled. Fig 3 shows the _ _ _
system in which machines in each area have been represented V 1 = jX 1 . I 1 + V Et (5)
by their Thevenin’s equivalent and L1, L2 are local loads in _ _ _ _
area 1 and area 2 respectively, UPFC is connected at mid- V Et =V B + jX 2b .I B +V 2 (6)
point. Although this is the simplest power system taken as an
where,
example, this will serve to illustrate the approach which
remains modular for larger system. It consists of shunt _ _
_
I 1 = I
_
E + I
_
B ;I 2 = − I B ; X 2b = X b + X 2
connected voltage source converters VSC-E via transformer
ET, a series connected voltage source converters VSC-B via Using (1) along with (5) and (6), d - and q - components of
transformer BT, both converters connected through dc link current I 1 and I 2 can be obtained as
capacitor C dc . In developing the model, UPFC based on
⎡X + X ⎤ ' ⎡ X ⎤
I1d = ⎢ 2b e
⎥E1q − ⎢ 2b ⎥.kmeVdc sinδe
[Δ δ Δ ω1 Δω2 Δ E1' q Δ Efd 1 Δ E 2' q Δ Efd 2 Δ V dc ]
T

⎢⎣ Xdee ⎥⎦ ⎣ Xdee ⎦
(7) Δu = [Δme Δδ e Δmb Δδ b ]T
;
⎡ X ⎤
[
− ⎢ e ⎥. kmbVdc sinδb + E2' q.cosδ − I2d .X2' d ] where, state matrix A and control input matrix B can be
⎣ Xdee⎦ evaluated.
The simplified Phillips-Heffron model for area 1 for UPFC
⎡X ⎤ ⎡ X ⎤ has been shown in Fig. 4. Similar model can be obtained for
I1q = ⎢ 2b ⎥.kmeVdc cosδe + ⎢ e ⎥.kmbVdc cosδb area 2 as well except the change in constants and sources. This
⎢⎣ X qee ⎥⎦ ⎢⎣ X qee ⎥⎦ can further be extended to multi area network with certain
(8) augmentation in the state matrix; however, the control input
⎡ X ⎤
[
+ ⎢ e ⎥. E2' q. sin δ + I2q.X 2q
⎢⎣ X qee ⎥⎦
] matrix will change with controller configuration selected.

A=
⎡ 0 ω0 − ω0 0 0 0 0 0 ⎤
I 2d
⎡ X ⎤
[
= ⎢ dt ⎥ . km b V dc sin δ b + E 2' q . cos δ − I 2 d . X '
2d ] ⎢

K11

D1
0
K 211
0
K 221
0
K pd1 ⎥

⎣ X dee ⎦ (9) ⎢
⎢ K
M1 M1 M1 M1 M1
K pd 2


12 D2 K 212 K 222
⎢ 0 − 0 0 ⎥
⎡ Xe ⎤ ' ⎡ X de ⎤ ⎢ M2 M2 M2 M2 M2 ⎥
−⎢ ⎥ . E 1q − ⎢ ⎥ .km e V dc sin δ e ⎢ K
41 K 311 1 K 321 K qd1 ⎥
⎣ X dee ⎦ ⎣ X dee ⎦ ⎢
⎢ Td' 01
0 0
Td' 01 Td' 01 Td' 01
0
Td' 01


⎢−K K − K a1K 611 −1 − K a1K 621 − K a1K vd1 ⎥
⎢ a1 51 0 0 0 ⎥
⎡ X qe ⎤ ⎡ X qt ⎤ ⎢ Ta1 Ta1 Ta1 Ta1 Ta1 ⎥
I 2q = ⎢ ⎥.km eV dc cos δ e − ⎢ ⎥.km bV dc cos δ b ⎢ K
⎢ 42 0 0
K 312
0
K 322 1 K qd 2 ⎥

⎢⎣ X qee ⎥⎦ ⎢⎣ X qee ⎥⎦ ⎢ Td' 02 Td' 02 Td' 02 Td' 02 Td' 02 ⎥
(10) ⎢− K K
⎢ a 2 52 − K a 2 K 612 − K a 2 K 622 −1 − K a 2 K vd 2 ⎥

[ ]
0 0 0
⎡ X qt ⎤ ' ⎢ Ta 2 Ta 2 Ta 2 Ta 2 Ta 2 ⎥
−⎢ ⎥. E 2 q . sin δ + I 2 q . X 2 q ⎢⎣ K7 0 0 K 81 0 K 82 0 − K9 ⎥⎦

⎢⎣ X qee ⎥⎦
and
where,
⎡ 0 0 0 0 ⎤
δ = δ 1 − δ 2 ; X qe = X 1q + X 1 ; X de = X '
1d + X 1
; ⎢ K pe 1 K pde 1 K pb 1 K pdb 1 ⎥
⎢ ⎥
⎢ M1 M1 M1 M1 ⎥
X = X + X ; X dt = X + X ⎢ K pe 2 K pd 2 K pb 2 K pdb 2 ⎥
qt qe e de e ⎢ ⎥
⎢ M 2 M 2 M 2 M 2 ⎥
X qee = X qe .X e + X qt .X 2b
; X dee = X de . X e + X dt . X 2 b ⎢ K qe 1 K qde 1 K qb 1 K qdb 1 ⎥
⎢ ⎥
T d' 01 T d' 01 T d' 01 T d' 01
B = ⎢ ⎥
The linearized equations from which the Phillips-Heffron ⎢ K a 1 K ve 1 K a 1 K vde 1 K a 1 K vb 1 K a 1 K vdb 1 ⎥
⎢ − − − − ⎥
model of an n-machine power network without FACTS ⎢ T a1 Ta1 T a1 Ta1 ⎥
⎢ K qe 2 K qde 2 K qb 2 K qdb 2 ⎥
devices is derived are [11-13] ⎢ ⎥
⎢ T d' 02 T d' 02 T d' 02 T d' 02 ⎥
⎢ K a 2 K ve 2 K a 2 K vde 2 K a 2 K vb 2 K a 2 K vdb 2 ⎥
⎢− − − − ⎥
⎢ Ta 2 Ta 2 Ta 2 Ta 2 ⎥
⋅ ⎫ ⎢⎣ K ce K cde K cb K cdb ⎥⎦
Δδ = ω 0 Δω ⎪
⋅ − ΔP − D .Δω ⎪ State-space model (12) has been obtained with UPFC.
Δω = e

M ⎪
⎪ (11)

' − ΔE 'q − ( X d − X '
d ) ΔI d + ΔE fd ⎬
ΔE q = ⎪
T d' 0 ⎪
⋅ ⎪
' − ΔE fd + K a ( − ΔV t ) ⎪
ΔE fd =
Ta ⎪⎭
The bold written variables with prefix Δ are all n-order
vectors and others are n-order diagonal matrices. The
complete state space model of two-area power network
installed with UPFC can be derived by combining and
linearizing eqs. (4) and (7) to (11) around an operating point.
The complete state space model of two-area power network
installed with UPFC can be obtained as
. Fig. 4. Modified Phillips-Heffron model of area 1 machine of
Δ x = AΔx + BΔu (12) two-area power network installed with UPFC
where,
In Fig. 2, the following vectors
Δx =
K p1 u , K , K v 1 u and K cu transfer the designer’s iteration on pole locations as used in
q1u
full state feedback to iterations on the elements in a cost
are defined as function, J. This method determines the feedback gain matrix
that minimizes J in order to achieve some compromise
K p1u = [K pb 1 K pdb 1 ], K [
q 1 u = K qb 1 K qdb 1 ], between the use of control effort, the magnitude, and the speed
K v1u = [K vb 1 K vdb 1 ] of response that will guarantee a stable system.
For a given system:
and K cu = [K cb K cdb ].
= Ax + Bu (13)

III. STATE PREDOMINANT MULTI-STAGE UPFC CCONTROL Determine the matrix K of the LQR vector
DESIGN U(t) = - K x (t) (14)
The dynamic characteristics of a system can be influenced So in order to minimize the performance index
by location of eigenvalues, for a good system response in

terms of overshoot/undershoot and settling time, a particular xT Q x U T R U dt (15)
location for system eigenvalues is desired depending upon the
operating conditions of system. The damping power and the where, Q and R are positive-definite Hermitian or real
synchronizing power are related respectively, to real part and symmetric matrix. Note that second term on the right side
imaginary part of eigenvalue that correspond to incremental account for the expenditure of the energy on the control
change in the deviation of rotor speed and the deviation of efforts. The matrix Q and R determine the relative importance
rotor angle[14], this eigenvalue is known as electromechanical of the error and the expenditure of this energy.
mode. Power oscillation damping can be improved if real part From the above equations,
of eigenvalue associated with mode of oscillation can be ∞
xT Q x x T K T R K x dt
shifted to left-side in complex s-plane as desired.
In multi-input-multi-output (MIMO) systems some states ∞ T
= x Q K T RK x dt (16)
are vulnerable and these states can be identified by the closed
loop poles of the system, i.e., system eigen values and For solving the parameter-optimization problem, Set
vulnerable states are inter-linked. Such deviated states may
x T (Q + KTRK) x
introduce instability which, in turn, deteriorates performance
of the system. The state predominant approach is a novel =-d/dt(xTPx) (17)
control approach by which the vulnerable states are regulated
Then it is obtained as,
with adequate control decisions by way of assessing the most
dominant eigen values and generating new control by xT (Q + KTRK) x = - xT P x – xT P x
modifying elements of Q matrix diagonal entry of respective
state and afterwards calculating feedback gains afresh for = - xT [(A-BK)T P +P (A-BK)] x (18)
ensuring system stabilization. In general, the concept of Comparing both sides of the above equation and note that this
assessing gain K is simple for small perturbations; however, equation must hold true for any x, it require that
for large and dynamically changing systems, the design based
on state predominant approach may be extended with multiple (A-BK)T P +P (A-BK) = - (Q+KTRK) (19)
iterations by evaluating eigenvalues along with analysis of Since R has been assumed to be a positive-definite Hermitian
their real and imaginary part for determination of impact on or real symmetric matrix as,
system instability. This may be done first with real part of R = TTT, Where, T is a non-singular matrix; and
eigen values very close to the imaginary plane and redefining
the Q matrix of concerned state diagonal entry, club such ATP + PA + [TK-(TT)-1BTP]T[TK-(TT)-1BTP]
nearest group of eigen values that affect the stability most -PBR-1BTP+Q=0 (20)
predominantly and thus corresponding diagonal entry may be
changed in an iterative mode. Such identified states are The minimization of J with respect to K requires the
termed as predominant states and regulation of heavily minimization of
deviated states is ensured by adequate strengthening of signal xT[TK-(TT)-1BTP]T[TK-(TT)-1BTP]x (21)
by assigning large weight in respective diagonal entries of Q
matrix. In general, R is maintained as Identity matrix. The Which is nonnegative, the minimum occurs when it is zero, or
multi-stage LQR uses state predominant concept in which the when
system eigen values of the most deviated states are taken as TK= (TT)-1BTP (22)
initial set of inputs for assigning the weights of concerned -1 T -1 T -1 T
diagonal entries of Q matrix, again a new set of gain array K Hence, K=T (T ) B P=R B P (23)
can be calculated to re-establish new state feedback in next control law,
stage and this process has to be repeated till all deviated states
are tracked to the desired criterion. U(t) = - K x (t) = -R-1 BT P x (t) (24)
One of the MIMO design approach is the optimal control P must satisfy the reduced Riccati equation and can be written
method of linear quadratic regulator (LQR). The idea is to as
ATP+PA – PBR-1BTP +Q = 0 (25) k3=lqr(A3, B, Q3, R)
A. Design of Multistage LQR Controller
5) Step V
This section deals with different design stages of LQR. A4 =( A3 - ( B *k3))
They are given as S4=eig(A4)
1) Design of Single -Stage LQR Controller K4=lqr(A4, B, Q4, R)
Single stage LQR controller has the following steps
6) Response calculation
(i) Obtain the array of gain K from initial system knowledge sys= ss(A4, B, C11, D11)
and some initialized Q matrix and R as Identity matrix.
(ii) A new state matrix A1 is obtained as A1= (A-(B*K), the This gives response of system to a step input. The multi-stage
eigen values of A1 and hence associated states which are can be decided based on the design considerations and
responsible for oscillations are identified and acceptability for a given system.
corresponding changes are made in the diagonal entry of
Q and step (i) is repeated to get a new set of gain array IV. CASE STUDY
and then step (ii) is evaluated for oscillatory states if any. The sample system considered has been analyzed with
This may subsequently lead to second -stage LQR. conventional LQR and multi-stage LQR and the results are
2) Design of Third- Stage LQR Controller tabulated as shown in Table I. Some of the simulation output
The gain matrix K1 obtained from K1= lqr (A1,B,Q1,R) is for a step change has been shown for δ & 11 in Fig. 5 (a, b, c
multiplied to the B matrix and a new system matrix is formed and d). The detailed variables have been listed in Table I.
which is a combination of plant and the controller and A1 is
transformed into A2=(A1-(B*K1) and the eigen values which
creates oscillations are identified and for that eigen values
selection of gain matrix Q1 should have higher gain and eigen
values obtained will be negative and vulnerability of the states
will be reduced followed by improved system performance
compared to second- stage LQR
3) Design of Multi-Stage LQR Controller
The gain matrix K2 obtained from K2=lqr(A2 ,B,Q2,R) is
multiplied to the B matrix . A2 is transformed into A3=(A2-
(B*K2)and eigen values which creates oscillations are
identified. This process will be same as given in 2) and will Fig. 5(a). Variation in δ (delta) with LQR
continue depending on the required specifications.
B. Implementation of Multi-stage LQR
The implementation of multi-stage LQR has been done by
using MATLAB functions. The system matrices obtained with
state space model has been used to first calculate eigenvalues
of system. The feedback gains are calculated by lqr function.
Initial weights of Q are assumed based on eigenvalues
calculated in first step.

1) Step I
S= eig(A)
K= lqr (A, B, Q, R)
Fig. 5(b). Variation in δ (delta) with multi-stage LQR
2) Step II
A1=( A - ( B *K))
S1=eig(A1)
k1=lqr(A1, B, Q1, R)

3) Step III
A2 =( A1- ( B *k1))
S2=eig(A2)
k2=lqr(A2, B, Q2, R)

4) Step IV
A3 =( A2 - ( B *k2))
S3=eig(A3) Fig.5 (c). Variation in 11 with LQR
VI. APPENDIX
The system data, with two machines with UPFC, used for
study has been given in Table II
TABLE II

Transformer: Xte =0.03, Xb= 0.30


Transmission line: Xe = 0.3, Xbv =0.03
Operating condition: Vb =1.0 p.u., δ= 15°
UPFC Parameter: me=0.85, δe=0.03, mb=0.85, δb=0.3
DC link capacitor: Cdc = 0.0005 p.u., Vdc0= 1.0 p.u.
Generator 1: M1= 0.0255 MJ/MVA, d1=0.0 sec., T'1d0=5.044 sec.
Fig. 5(d). Variation in 11 with multi-stage LQR X1d=.190, X1q=0.163, X1d1=0.0765, E1q1=1.024 p.u.
Excitation system 1: K1A=10; T1A=0.010 sec.
Generator 2: M2=0.0255MJ/MVA, d2=0.0 sec., T'2d0=5.6 sec.,
TABLE I X2d=.190, X2q=.163, X2d1=0.0765, E2q1=1.0 p.u.
COMPARATIVE STUDY OF CONVENTIONAL & MULTI-STAGE Excitation system 2: K2A=20, T2A=0.010sec
LQR UPFC
VII. REFERENCES
[1] Larsen, E.V., Sanchez-Gasca, J.J., Chow J. H.: “Concept for Design
of FACTS Controllers to Damp Power Swings”, IEEE Trans. on Power
Syst., Vol. 10, No.2, pp. 948-956, 1995.
[2] Noroozian M., Anderson G.: “Damping of Power System Oscillations
by use of Controllable Components”, IEEE Trans. on Power Del., vol.9,
No.4, pp. 2046-2054, 1994.
[3] L. Gyugyi, “Dynamic compensation of AC transmission line by solid-
state synchronous voltage sources”, IEEE Trans. Power Del., vol. 9, No.
2, pp. 904-911, 1994.
[4] Higorani, N.G., and Gyugyi, L., Understanding FACTS: Concepts and
Technology of Flexible AC Transmission Systems, IEEE Press, 1999.
[5] L. Gyugyi “A Unified Power Flow Control Concept of Flexible AC
Transmission Systems”, IEE Proceedings – C, Vol. 139, No. 4, pp. 323-
331, July 1992.
[6] L. Gyugyi, C. D. Schauder, S. L. Williams, T. R. Rietman, D. R.
Torgerson, and A. Edris “The Unified Power Flow Controller: A New
Approach to Power Transmission Control”, IEEE Trans. on Power Del.,
Vol. 10, No. 2, pp. 1085 – 1093, April 1995.
[7] H. F. Wang and F. J. Swift, “A Unified Model for the Analysis of
FACTS Devices in Damping Power System Oscillations Part III:
Unified Power Flow Controller,” IEEE Trans. on Power Del., vol. 15,
No. 3, pp. 978- 983, July 2000.
[8] T. K. Mok, Y. Ni and F.F. Wu, “Design of Fuzzy Damping Controller
V. CONCLUSIONS of UPFC through Genetic Algorithm”, IEEE Power Engineering Society
Summer Meeting, Vol. 3, pp.1889 – 1894, 2000.
A multi-stage LQR concept introduced in this paper, which [9] L. Y. Dong, L. Zhang and M.L. Crow, “A New Control Strategy for the
is based on state predominant concept, gives settling time in Unified Power Flow Controller”, IEEE Power Engineering Society
fraction of seconds with minimum overshoots for multi Winter Meeting, Vol. 1, pp.27-31, 2002.
machine system. For multi machine system some of the states [10] Z. Huang, Y. Ni, C.M. Shen, F.F. Wu, S. Chen and B. Zhang,
“Application of Unified Power Flow Controller in Interconnected Power
having larger settling time with conventional LQR are well Systems – Modeling, Interface, Control Strategy, and Case Study”,
regulated with multi-stage LQR in terms of settling time IEEE Trans. on Power Syst., Vol. 15, No. 2, pp. 817-823, 2000,
reduction for the same perturbation and loading. The concept [11] Yu Y. N, “Electric Power System Dynamics”, Academic Press-1983.
of multi area interconnected power network oscillations has [12] DeMello F. P., Concordia C., “Concept of Synchronous Machine
Stability as Affected by Excitation Control”, IEEE Trans. PAS, Vol. 88,
been thoroughly examined in the true perspective of damping.
pp. 316-329, April 1969.
During the course of investigation it has been observed that [13] Heffron W.G., Phillips R.A. “Effect of Modern Amplidyne Voltage
classical control of inter area oscillations may not give fastest Regulator on Under Excited Operation of Large Turbine Generator”,
possible time response. The sample power system has been AIEE Trans, Vol. 71, pp. 692-697, 1952.
[14] P. Kundur, Power System Stability and Control, McGraw Hill, 1994.
studied with both LQR and multi-stage LQR control strategy, New York.
where the results for later case give optimal gains for tracking [15] L. S. Shieh, H. M. Dib and B. C. Mcinnis, “Linear Quadratic Regulators
system dynamics adequately. Moreover, a concept of defining with Eigenvalue Placement in a Vertical Strip”, IEEE Trans. on
intelligent weight matrix has been introduced with proposed Automatic Control, Vol. AC-31, pp. 241-243,1986.
[16] R. V. Patel, Neil Munro, Multivariable System Theory and Design, Int.
state predominant concept for state regulation in general. The Series on System and Control, Vol. 4, Pergamon Press, New York
more deviated state, the larger diagonal weights may be [17] R. K. Pandey and N. K. Singh, “Small Signal Model for Analysis and
attached, thus giving maximum possible fastest settling time. Design of FACTS Controllers”, IEEE PES General Meeting 2009

View publication stats

You might also like