Documente Academic
Documente Profesional
Documente Cultură
7, JULY 2009
2493
AbstractThis paper presents a solution to the control of a unified power quality (PQ) conditioner (UPQC) for PQ improvement
in power distribution systems. The problem formulation allows
not only harmonic compensation but also voltage sags/swells,
load demand changes, and power factor correction to be tackled
in a unified framework. The proposed controller combines the
multivariable regulator theory with H loop shaping, so that
zero steady-state error, robustness to modeling uncertainties, and
insensitivity to supply frequency variations can be accomplished
simultaneously, thus providing a complete theoretical solution to
all the aforementioned PQ problems. The effectiveness of the
proposed controller is, in practice, verified by experimental studies
on a single-phase power distribution system.
Index TermsActive filter, harmonic compensation, H-infinity
loop shaping, Kalman filters, multivariable regulator, power quality (PQ), unified PQ conditioners (UPQCs).
I. I NTRODUCTION
Manuscript received July 16, 2008; revised November 18, 2008 and
February 18, 2009. First published April 21, 2009; current version published
July 1, 2009. This work was supported by the School of Electrical and
Electronic Engineering, Nanyang Technological University, Singapore.
K. H. Kwan is with the Laboratory for Clean Energy Research, School
of Electrical and Electronic Engineering, Nanyang Technological University,
Singapore 639798, and also with Temasek Polytechnic, Singapore 529757
(e-mail: kwan0019@ntu.edu.sg).
Y. C. Chu is with the Division of Control and Instrumentation, School
of Electrical and Electronic Engineering, Nanyang Technological University,
Singapore 639798.
P. L. So is with the Laboratory for Clean Energy Research, School of Electrical and Electronic Engineering, Nanyang Technological University, Singapore
639798.
Digital Object Identifier 10.1109/TIE.2009.2020705
injecting distorted components, e.g., harmonics (from sampled supply voltage and load current), into the network [20]
[24]. This aims to make the load voltage and supply current
undistorted. However, there is no feedback (FB) of the load
voltage and supply current in these designs to show that they
are undistorted. From this point of view, these designs are
open-loop control. Also, the shunt and series filters of these
approaches are usually controlled independently, despite the
existence of a coupling effect between the series and shunt
filters [25]. Furthermore, if there is a drift in the operating
frequency, the extraction of harmonic components can become
inaccurate. For these reasons, this type of control approach
might not result in the best performance of the UPQC.
Another approach is the model-based control [26], [27].
Here, the UPQC is modeled with a coordinated control scheme.
The coupling effect is considered in the UPQC model. However, the cost and complexity of the UPQC will increase due
to the additional sensors or high-speed DSP needed. Moreover,
any modeling error can be detrimental to its performance.
The proposed approach in this paper is a model-based feedforward (FF)/FB control. First, a UPQC model is formulated.
Then, the FF control is applied to make the model outputs, i.e.,
the load voltage and the supply current, track certain desired
waveforms. It can ensure zero steady-state tracking error if
there is no modeling error. Otherwise, the negative effects
introduced by any modeling error can be mitigated by the
FB control simultaneously. This approach provides direct FB
to the controller and shows how well the PQ of the outputs
can be improved. The coupling effect between the series and
shunt filters is also taken care of. To minimize the cost, digital
Kalman filters are implemented in place of hardware sensors.
To minimize the complexity of the UPQC, the gains of the
controller and the Kalman filters are calculated offline and
then implemented as constant gains online. Also, the proposed
control approach ensures that the UPQC can still operate under
slight frequency variations in the supply voltage.
In what follows, this paper provides a complete solution for
the operation of a UPQC from the control theoretic point of
view, based on a multi-inputmulti-output (MIMO) state-space
model that is to be presented in Section III. The following four
tasks will be accomplished simultaneously by the UPQC:
1) compensating the harmonics in the supply voltage and
load current;
2) eliminating the disturbances due to voltage sags/swells at
the supply side or changes in the load demand;
3) correcting the power factor at the supply side;
2494
Fig. 1. Classical control approach for the series active filter of the UPQC.
Fig. 2.
2495
l
R
Ll
0
A= 0
1
C
0
se
R
Lse
0
1
Cse
se
1
Csh
1
Ll
D1 =
0
1
0
0
0
0
B1 =
0
0
0
C=
C1sh
0
0
0 0
0 0
0
0
0
0
sh
R
Lsh
0
1
Csh
L1l
L1se
0
0
0
Vdc
2L
se
B2 = 0
0
L1l
0
L1sh
0
0
0
0
Vdc
2Lsh
0
0
1
0
0 0
D2 =
0 0
0
0
(3)
The first step of our control design is to represent the periodic disturbance of the model [vS iL ]T by another statespace model known as the exosystem. The basic idea is that
any periodic signal v(t)
(4)
v = C
(5)
2496
= A + L w
(6)
v = C .
(7)
vf dvL ,
0,
if no voltage sag/swell
if voltage sag/swell.
(8)
|
vf ||diS | = |dvL ||if | cos
(9)
(10)
= A + L w
vS
iL = C1
(11)
(12)
d = C2
(13)
(14)
C2 = CX + B0 + D2 U
(15)
D0 = D1 C1 .
(16)
u = U .
(17)
X w + B2 (u U )
(x X )
dt
+D
1 w + D2 (u U )
e = C(x X )
(18)
(19)
where
X = B
1 XL ,
B
1 = [B1
B
0],
1 = [D1
D
0]
e = y d.
(20)
Fig. 3.
2497
FF control for the UPQC, with K/F denoting the Kalman filter.
Fig. 5. Magnitude plot of W2 (s) with three peaks at 50, 150, and 250 Hz.
Fig. 4. Combining the precompensator W1 (s), the postcompensator W2 (s),
and the stabilizing controller K (s) into an overall FB controller.
series in the formulation of the exosystem. Therefore, a complimentary FB control, which can be implemented easily, is
designed to mitigate the negative effects of any modeling errors
and to ensure satisfactory performance of the UPQC under
slight drift of the fundamental frequency.
B. Design of FB Control
In this paper, the FB control law is designed based on the H
loop shaping.
For a given linear time-invariant MIMO plant G(s), the idea
of H loop shaping [32][34] is to employ a precompensator
W1 (s) and a postcompensator W2 (s) to modify the open-loop
gain to a desired one and then close the loop with a stabilizing
controller K (s) that is computed by some H optimizations
to maximize the stability margin. Thus, the overall FB controller is given by W1 (s)K (s)W2 (s), as shown in Fig. 4.
Although the synthesis of the stabilizing controller K (s)
is pretty systematic, the design of the precompensator W1 (s)
and the postcompensator W2 (s) is problem dependent and
requires some engineering skills. Several case studies of H
loop shaping in power system applications can be found in
[40][42]. In [40], where H loop shaping was applied to
the steam-generator-level control in EDF nuclear power plants,
the precompensator W1 (s) was chosen as a second-order lowpass filter to increase the roll-off rate at high frequencies for
better noise attenuation, whereas the postcompensator W2 (s)
was chosen as a PI controller for zero steady-state error. On the
other hand, for the power system stabilizers in [42], W1 (s) was
simply set to one, whereas W2 (s) was designed to increase the
loop gain in the frequency range of the interarea mode but not
at other frequencies.
Our H loop-shaping approach to the UPQC is motivated
by the idea that slight frequency variations can be tackled by
properly designing W1 (s) and W2 (s), while the impedance
variations in the system can be handled by the robust stabilizing
controller K (s). Note that the values of the line impedance
and VSI impedances are not exactly known in practice. Therefore, the robustness offered by K (s) is an effective means
to maintain a desirable performance under such modeling
uncertainties.
To design W1 (s) and W2 (s), we note that the signals considered here are periodic. Therefore, the desired open-loop
(s + i )2
s2 + 2i s + i2
i=1
(21)
(co /1 )2 + (co /1 )2 2
2(co /1 )2 + 2(co /1 )2
(22)
s + 101
10s + 1
(23)
2498
(2) to become
Fig. 6. FB control for the UPQC, with FB/C denoting the FB controller.
(24)
L
0
A 0
u
+ w+
=
B0 A x
B2
x
B1
0
d
0
C2 0
u
=
+ w+
D0 C x
D2
y
D1
(26)
(27)
(25)
Both computer simulation and experiment have been conducted to verify the control performance of the UPQC, but
only the experimental results are reported in this paper for
brevity. A single-phase power distribution system supplied by
a programmable voltage source, rated at 100 VRMS 50 Hz, is
constructed. Several light dimmers are used as the nonlinear
load. By adjusting the light intensity of the dimmers, the
demand and the harmonic content of the load current iL can
be varied. The supply voltage vS at the PCC can also be
distorted to a variable degree by paralleling additional light
dimmers. Note that a single-phase system is used merely for
simplicity. Our control methodology can easily be extended to a
Fig. 8.
2499
Overall configuration of the UPQC, with K/F denoting the Kalman filter and FB/C denoting the FB controller.
TABLE I
VALUES OF THE COMPONENTS OF THE UPQC
what follows, four test cases are carried out to investigate the
performance of the UPQC.
2500
2501
Fig. 11. Experimental test case 2: rms value of voltages and currents.
Fig. 12. Experimental test case 2: power delivered by (solid lines) the supply and consumed by (dashed lines) the load.
2502
Fig. 13. Experimental test case 3: (top) load current waveforms and (bottom) supply current waveforms after compensation.
Fig. 15. Experimental test case 4: harmonic spectrum of voltages and currents when vS is at 47 Hz.
TABLE III
BREAKDOWN OF THD VALUES (IN PERCENT) OF VOLTAGES AND
CURRENTS UP TO THE 17TH ORDER
TABLE IV
POWER QUALITY FLUCTUATIONS DURING FREQUENCY VARIATIONS
steady-state error in principle, and H loop shaping to synthesize an FB controller that robustly stabilizes the closed-loop
system. By properly choosing the innovations of the Kalman
filters and the loop shape, a wide range of power quality issues
have been put into the same methodological framework and
tackled simultaneously. Experimental studies have confirmed
that the H loop-shaping design is robust to modeling errors,
including the nonlinear effects introduced by the PWM and
the uncertain values of the impedance in the line and VSIs.
Furthermore, the proposed loop shape is an effective means
to maintain a good control performance when slight frequency
variations occur in the supply voltage. It is concluded that this
paper has demonstrated a systematic approach to the control
design of UPQC, providing an overall solution to a variety of
power quality problems encountered in a power distribution
system.
A PPENDIX
To derive a state-space model for the UPQC in Section III,
Kirchhoffs voltage and current laws are applied to the three
current loops shown in Fig. 2.
From loop iS , we obtain
vS = iS Rl + Ll
diS
+ vinj + vL .
dt
(28)
Vdc
dise
= Rse ise + Lse
+ vinj
2
dt
1
(iS + ise )dt.
vinj =
Cse
(29)
(30)
u2
iS + iinj iCsh = iL .
(31)
(32)
(33)
2503
2504
[31] S. Hara, Y. Yamamoto, T. Omata, and M. Nakano, Repetitive control system: A new type of servo system for periodic exogenous
signals, IEEE Trans. Autom. Control, vol. 33, no. 7, pp. 659668,
Jul. 1988.
[32] D. C. MaFarlane and K. Glover, Robust Controller Design Using Normalized Coprime Factor Plant Descriptions, vol. 138. Berlin, Germany:
Springer-Verlag, 1990.
[33] D. C. MaFarlane and K. Glover, A loop shaping design procedure using
H8 synthesis, IEEE Trans. Autom. Control, vol. 37, no. 6, pp. 759769,
Jun. 1992.
[34] G. Vinnicombe, Uncertainty and Feedback: H8 Loop-Shaping and the
V-Gap Metric. London, U.K.: Imperial College Press, 2001.
[35] A. Ghosh and G. Ledwich, Power Quality Enhancement Using Custom
Power Devices. Norwell, MA: Kluwer, 2002, pp. 380406.
[36] K. H. Kwan, P. L. So, and Y. C. Chu, Unified power quality conditioner
for improving power quality using MVR with Kalman filters, in Proc.
Int. Power Eng. Conf., 2005, vol. 2, pp. 980985.
[37] J. A. Macias and A. Gomez, Self-tuning of Kalman filters for harmonic
computation, IEEE Trans. Power Del., vol. 21, no. 1, pp. 501503,
Jan. 2006.
[38] V. Moreno, A. Pigazo, and R. I. Diego, Reference estimation technique
for active power filters using a digital Kalman algorithm, in Proc. Int.
Conf. Harmonics Quality Power, 2002, vol. 2, pp. 490494.
[39] S. Liu, An adaptive Kalman filter for dynamic estimation of harmonic
signals, in Proc. Int. Conf. Harmonics Quality Power, 1998, vol. 2,
pp. 636640.
[40] P. Ambos, G. Duc, and C. M. Falinower, Loop shaping H8 design
applied to the steam generator level control in EDF nuclear power plants,
in Proc. IEEE Int. Conf. Control Appl., 1996, pp. 751756.
[41] R. F. Pannett, P. K. Chawdhry, and C. R. Burrows, Alternative robust
control strategies for disturbance rejection in fluid power systems, in
Proc. Amer. Control Conf., 1999, pp. 739743.
[42] C. Zhu, M. Khammash, V. Vittal, and W. Qiu, Robust power system
stabilizer design using loop shaping approach, IEEE Trans. Power Syst.,
vol. 18, no. 2, pp. 810818, May 2003.
[43] A. Emadi, A. Nasiri, and S. B. Bekiarov, Uninterruptible Power Supplies
and Active Filters. Boca Raton, FL: CRC Press, 2005, pp. 117131.