Abstract In this paper, a control strategy of the single-

phase active filter using a novel d-q transformation is pro-

posed. This d-q transformation is implemented by using
Hilbert transform, by which the instantaneous single-
phase voltage and current are converted into complex vec-
tors (analytic signals) on the instantaneous basis. Since
the fundamental components of voltages and currents are
converted into dc components on the d-q coordinate both
in the steady and transient states, the harmonics of volt-
ages and currents can be obtained precisely through the
low-pass filer installed on the d-q coordinate. This results
in the economical active filter with a small capacitor sup-
pressing the dc bus voltage fluctuation because the fun-
damental power can be fully removed from the active fil-
ter input.
Keywords: Single-phase, Active filter, Instantaneous ac-
tive power and reactive power, Hilbert Transform.
I. INTRODUCTION
For suppression of harmonic voltages and currents in power
distribution systems, various power active filters (AFs) have
been proposed [1]-[5]. The most important technology on
AFs is the detecting method of harmonics to reduce the ca-
pacity of the energy storage component. Akagi et al. intro-
duced the theory of instantaneous reactive power for the three-
phase circuit to determine harmonics from non-sinusoidal
voltages and currents[1]. Their theory can be reformulated
for the three-phase four wires system [2] and other multi
phase systems [3]. However, it can not be applied to the single-
phase circuit, because the instantaneous reactive power in
their theory is defined as the power flowing from one phase
to the other.
Generally, harmonics in single-phase voltages and currents
are calculated based on the distorted wave circuit theory, such
as the Fourier analysis, in which the repeated waveforms of
voltages and currents are considered. Therefore, harmonics
in transient states are not defined precisely. In the practical
single-phase active filter, by this problem, the output volt-
age and current tend to contain a large amount of fundamen-
A Control Strategy of Single-phase Active Filter
Using a Novel d-q Transformation
Makoto Saitou, Nobuyuki Matsui
Dep. of Electrical and Computer Engineering
Nagoya Institute of Technology
Gokiso, Showa, Nagoya, Aichi, JAPAN
Toshihisa Shimizu
Dep. of Electrical Engineering
Tokyo Metropolitan University
1-1 Minami-ohsawa, Hachiohji, Tokyo, JAPAN
tal components, and the dc bus voltage tend to be fluctuated
intensely for the non-periodic load changes. Then, the large
size of the chemical capacitor is needed at the dc bus of the
active filter so as to avoid this fluctuation. However, from
the economical point of view, the large size of the chemical
capacitor should not be installed, because this capacitor re-
quires the large space and its lifetime is shorter than the
other type of capacitors.
In this paper, the harmonics detecting method for single-
phase voltages and currents using a novel d-q transforma-
tion is proposed. In this method, the instantaneous single-
phase voltage and current are converted into the complex
vectors (analytic signals) by Hilbert Transform on the in-
stantaneous basis [6], and these complex vectors are trans-
formed into the d-q axis components. Through this d-q trans-
formation, single-phase instantaneous active power and re-
active power are defined[7][8], and harmonic components of
single-phase voltages and currents are calculated by AC com-
ponents of these powers. In addition, the detecting method
of the instantaneous phase angle of a single-phase voltage is
proposed.
The significant characteristic that the fluctuation of the dc
bus voltage of the active filter can be suppressed without us-
ing the large size of a dc bus capacitor is verified through
the 100V/1.5kVA experimental system.
II. SINGLE-PHASE INSTANTANEOUS ACTIVE AND REACTIVE
POWER THEORY (ps-qs THEORY)
A. A single-phase d-q transformation using Hilbert Trans-
form
Generally, a real signal ( ) x t can be defined as
( )
( )
( ) ( )
0 0
0
t
x t
f t t
<

, (1)
where ( ) f t is an arbitrary real signal.
And the Hilbert Transform of ( ) x t is defined as
( )
( )
0
t P x
y t d
t

, (2)
where P is Cauchy principal value.
0-7803-7883-0/03/$17.00 2003 IEEE
1222
Fig.2 Vector diagrams on d-q coordinate.
Fig.1 Vector diagrams of a single-phase voltage and
current.
As a result, the analytic signal of ( ) x t is defined as
( ) ( ) ( ) z t x t jy t +

, (3)
Using Hilbert Transform, a single-phase voltage ( ) e t

and
current ( ) i t

can be represented as the analytic signals, which

are defined as complex vectors on the coordinate shown
in Fig.1 , as follows:
( ) ( ) ( ) e t e t je t

, (4)
( ) ( ) ( ) i t i t ji t

, (5)
where
( )
( )
0
t P e
e t d
t

, (6)
( )
( )
0
t P i
i t d
t

. (7)
A d-q transformation can be applied as following conven-
tional methods:
( ) ( )
( ) j t dq
e t e t e



, (8)
( ) ( )
( ) j t dq
i t i t e



, (9)
where ( ) t is a reference angle of the d-q coordi-
nate.
B. Definition of a single-phase instantaneous active power
and reactive power
Selecting a reference angle of the d-q coordinate ( ) t to
an instantaneous phase angle of a single-phase voltage ( ) e t

,
the vector diagrams based on (8) and (9) are depicted in Fig.2.
As apparent from this Figure, ( ) e t

and ( ) i t

can be ex-
pressed as
( ) ( ) ( ) cos
d
e t e t t

, (10)
( ) ( ) ( ) ( ) ( ) cos sin
d q
i t i t t i t t

. (11)
It is clear that ( ) i t

can be split into two components, one

is in phase with ( ) e t

and another is led ( ) e t

by exactly
90

. As a result, a single-phase active current ( )

pS
i t and
reactive current ( )
qS
i t are defined as
( ) ( ) ( ) cos
pS d
i t i t t
, (12)
( ) ( ) ( ) sin
qS q
i t i t t . (13)
Furthermore, a single-phase instantaneous active power ( )
S
p t
and reactive power ( )
S
q t are defined as follows:
( ) ( ) ( ) ( )
( ) ( ) ( )
cos
1
1 cos2
2
S d pS
d d
p t e t t i t
e t i t t

+
, (14)
( ) ( ) ( ) ( )
( ) ( ) ( )
cos
1
sin 2
2
S d qS
d q
q t e t t i t
e t i t t

. (15)
Supposing ( ) e t

and ( ) i t

are given as fundamental com-

ponents:
( ) ( )
1 1
cos e t E t

, (16)
( ) ( )
1 1
cos i t I t

+ , (17)
the d-q components in terms of ( ) e t

and ( ) i t

are defined
as the following dc components:
( )
( )
1
0
d
q
e t E
e t
, ] , ]

, ] , ]
] ]
, (18)
( )
( )
1
1
cos
sin
d
q
i t I
i t I

, ] , ]

, ] , ]

] ]
, (19)
and ( )
S
p t , ( )
S
q t are determined as:
( ) ( )
1 1 1
1
cos 1 cos 2
2
S
p t E I t +
, (20)
( ) ( )
1 1 1
1
sin sin2
2
S
q t E I t
. (21)
In this case, ( )
S
p t and ( )
S
q t correspond to the single-phase
fundamental active power and reactive power, respectively.
III. HARMONIC DETECTION METHOD
From (18) and (19), it is clear that fundamental compo-
nents of ( ) e t

and ( ) i t

are converted into the dc compo-

nents through the d-q transformation. As a result, a harmonic
detection strategy for ( ) i t

can be obtained as shown in Fig.3.

( ) ( ) ( ) cos
d
e t e t t

( ) ( ) ( ) cos
pS d
i t i t t ( ) ( ) ( ) sin
qS q
i t i t t
( ) i t

( ) t
( ) t
( )
d
e t
( )
d
i t
( )
q
i t ( ) e t

( ) i t

d
q
j
j

( ) i t

( ) e t

( ) e t

( )
( )
0
t P e
e t d
t

( ) i t

( )
( )
0
t P i
i t d
t

( ) t
j
1223
Fig.3 Harmonic detection strategy
for a single-phase current.
Fig.4 dc bus voltage regulator. Fig.5 Block diagram of a dc bus voltage regulator.
(This strategy can be reformulated for a harmonic detection
strategy of ( ) e t

.)
( ) i t

is converted into the analytic signal ( ) ( ) i t ji t

+ by
Hilbert Transform, and this analytic signal is transformed to
the d-q components ( )
d
i t , ( )
q
i t with a reference angle ( )
1
t .
The dc components of ( )
d
i t , ( )
q
i t are the magnitudes of the
fundamental active and reactive current, respectively. Then,
these components are determined from ( )
d
i t , ( )
q
i t by using
the low-pass filters (LPFs), and these are converted into the
analytic signal ( ) ( )
1 1
i t ji t

+ . Since the imaginary part of the
analytic signal ( )
1
i t

is a fictitious parameter, ( )
1
i t

is ig-
nored.
As a result, the fundamental component of ( ) i t

is deter--
mined as ( )
1
i t

, and the harmonic component of ( ) i t

is de-
rived from ( ) ( ) ( )
1 h
i t i t i t

.
IV. DC BUS VOLTAGE REGULATION
As well-known, the dc bus voltage of the active filter is
regulated by controlling the fundamental active power drown
by the active filter. In the single-phase active filter, this power
is defined as:
( ) ( ) ( ) ( )
1 1
1
1 cos 2
2
SC d Cd
p t e t i t t +
, (22)
where ( )
1 Cd
i t is the d-axis component of the funda-
mental output current of the active filter.
The relationship between the dc bus voltage of the active
filter ( )
DC
v t and ( )
SC
p t are given as:
( )

( )
2
1
2
AF
SC DC
C d
p t v t
r dt
+
, (23)
where
AF
C is the dc bus capacitor and r is the
equivalent resistor of the power circuit of the ac-
tive filter.
It should be noticed that (23) is nonlinear, because ( )
SC
p t
is the function of ( )
1
cos 2 t . Then, ( )
SC
p t is split into the dc
component ( )
SC
p t and the ac component ( )
SC
p t
as:
( ) ( ) ( )
1
2
SC d dC
p t e t i t
, (24)
( ) ( ) ( )
1
cos 2
SC SC
p t p t t
. (25)
And ( )
SC
p t is controlled for dc bus voltage regulation.
Figure 4 shows the proposed dc voltage regulator. In this
strategy, ( )
2
DC
v t is feeded back and the reference active
power ( )
SC
p t

is calculated. From ( )
SC
p t

and ( )
d
e t , the d-
axis reference current ( )
1 Cd
i t

is derived. ( )
1 Cd
i t

and the q-
axis reference current ( )
1 Cq
i t

which is set to zero are trans-

formed to the analytic signal ( ) ( )
1 1 C C
i t ji t


+ . Since ( )
1 C
i t

is a fictitious parameter, ( )
1 C
i t

is only given as the refer--

ence output current for the dc bus voltage regulation.
Figure 5 shows the block diagram of the dc bus voltage
regulator. In this diagram, the s-domain function of ( )
SC
p t
is expressed as:
( ) ( ) ( )
1 1 SC SC SC
p s p s j p s j + +
, (26)
where
1
is the angular frequency of the funda-
mental component of ( ) e t

.
And ( )
SC
p s
acts as the disturbance power in the dc bus volt-
age regulator.
When the bandwidth of ( )
SC
p s is limited below
1
, ( )
SC
p s
is ignored, because ( )
1 SC
p s j is given as the negative fre-
quency component and ( )
1 SC
p s j + is diminished by the low-
pass filter LPF1.
As a result, the proposed dc bus voltage regulator is oper-
ated linearly under this bandwidth limitation.
V. DETECTION OF INSTANTANEOUS PHASE ANGLE OF LINE
VOLTAGE
In order to perform the foregoing d-q transformation, the
instantaneous phase angle of the fundamental single-phase
voltage has to be detected. Generally, PLL circuits are used
for this purpose, where these circuits are synchronized with
the zero-crossing of a voltage. But, this method has the big
settling time and lack of stability.
Figure 6 shows the proposed phase angle detector. This
method is operated based on the d-q transformation in (8)
Hilbert
Transformer
LPF
LPF
( ) i t

( ) i t

( )
1
j t
e

( )
1
j t
e

( )
1
i t

( )
1
i t

( )
h
i t
( )
d
i t
( )
q
i t
( )
d
i t
( )
q
i t
( )
1
t
( )
C
1 T
d
e s
s +
1 T
T
P
s
K
s
+
1
2
AF
r
rC
s +
( )
2*
DC
v s
( )
Cd
i s

( )
SC
p s
( ) ( ) 2 2
SC SC
p s j p s j + +
( )
SC
p s
f
1
1 T s +
LPF1 PI
( )
2
DC
v s
C
T : Time constant of AF output current control
( )
2
DC
v t

( )
2
DC
v t
PI
( )
1 Cd
i t

( )
1
0
Cq
i t

( )
1 C
i t

( )
1 C
i t

( )
1
t
( )
1
j t
e

LPF1
( )
1
2
d
e t
( )
CS
p t

1224
Fig. 6 Instantaneous phase angle detector.
Fig.7 Block diagram of an instantaneous
phase angle detector.
where the reference angle is given as:
( ) ( )
1
t dt t +

, (27)
where
( ) ( ) KI
q
t e t dt

, (28)
KI

is a gain for phase angle detection.

It is clear that the fundamental component of ( ) e t

is trans-
formed to the dc components by using this d-q transforma-
tion. Then, the following equation is found:
( ) ( )
( ) ( ) [ ( ) ( ) ]
1 1 1
cos sin
d q
e t je t
E t t j t t
+
+
, (29)
where ( )
d
e t , ( )
q
e t are the dc component of ( )
d
e t
and ( )
q
e t , respectively..
Introducing the approximation of ( ) ( )
1
sin t t into
Eq.(29), ( )
q
e t is reformulated as:
( ) ( ) ( )
( ) ( )
1 1
1 1 1
KI
q
q
e t E t t
E t dt e t dt

. (30)
From Eq.(30), the block diagram of the proposed phase
angle detector can be obtained as shown in Fig.7, in which
( )
q
e s
is the ac component of ( )
q
e s . As apparent from this
Figure, ( )
q
e s is controlled to zero as
( )
( )
1
0 0
1
KI
lim lim 0
KI
q
s s
q
e s
s E
s
e s s E

+
. (31)
Therefore, ( ) ( )
1
t t is obtained.
VI. IMPLEMENTATION OF HILBERT TRANSFORM
The circuit implementing the Hilbert Transform is called
Hilbert Transformer and its transfer function is defined as
( )
( )
( )
0
0
j
H
j

>

<

. (32)
In the practical system, the filter circuits, such as the FIR
filter, are used [9],[10] as Hilbert transformer. Almost all
the filer circuits have gain and phase errors. Therefore, it is
difficult to obtain the characteristic defined as Eq.(33) strictly.
In Fig.3, ( )
d
i t and ( )
q
i t are defined by the fundamental
component of ( ) i t

, hence, the harmonics of ( ) i t

does not
affect the determination of ( )
d
i t and ( )
q
i t practically. And
in Fig.6, ( )
q
e t is used for detection of the phase angle, that
is, the harmonics component of ( ) e t

can be ignored. Then,

the following function is used for the Hilbert Transformer in
Fig.3 and 6:
( )
1
1
s
H e

, (33)
This function can act as the ideal Hilbert Transformer only
at the angular frequency
1
, hence,
( )
1 1
H j
. (34)
VII. EXPERIMENTAL RESULTS
A. Configuration of test system
Figure 8 shows the configuration of the experimental sys-
tem and Table.1 gives the parameters of the system. In this
system, the active filter compensates for the harmonic com-
ponent of the load current which is drown by the capacitor-
input diode rectifier. The harmonic component of ( ) i t

is
determined based on Fig.3. The instantaneous phase angle
with respect to the fundamental component of the line volt-
age is detected based on Fig.6, and the dc bus voltage of the
active filter is regulated based on Fig.4. The output current
Hilbert
Transformer
( ) e t

( ) e t

( ) ( )
1
t t
( ) j t
e

( )
d
e t
( )
q
e t
KI dt

1
dt

( ) t
Fig. 8 Test system configuration.
e

S
i

i

C
i

DC
v
AF
C
LO
C
L
R
S
L
LO
L
AF
L
Fig.6
Fig.3
Fig.4
PI
i

h
i
C
i

100V
DC
v
1

1 C
i

KI
s

1
E
( ) s ( ) s
1
s ( )
1
s
( )
q
e s
( ) 0
q
e s

( )
q
e s
( )
q
e s
1225
240
250
260
v
DC
*
v
DC
[V]
i
Sa
[A]
i
ca
[A]
i
a
, i
ah
[A]
t
a
Time [12.5ms/div]
-20
0
20
-20
0
20
-30
0
30
of the active filter is controlled to its reference by PI control-
ler. All control procedures are implemented by a DSP sys-
tem.
B. Results of instantaneous phase angle detection
Figure 9 shows the experimental waveforms of the esti-
mated instantaneous phase angle of the line voltage . After
the estimation circuit is started, the instantaneous phase angle
can be calculated within 15[ms]. Regardless of the large dis-
tortion of the line voltage, there is no phase error in steady
states.
C. Results of harmonic compensation
Figure 8 shows the experimental waveforms of the har-
monic compensation. In this experiment, the stepwise load
change is given from 1.5 kW to 500W at time t
a
, therefore,
the load current is changed from 30 A (peak) to 10 A(peak)
at time t
a
. In this condition, the dc bus voltage of the AF is
controlled without a large amount of fluctuation in spite of
small dc bus capacitor. Furthermore, vibrations are not ob-
served in the compensation current and the supply current
both in the steady and transient states. After the proposed
AF is operated, THD of the supply current is reduced from
33.78 % to 3.70 %, and harmonics components are reduced
as shown in table.2.
VIII. CONCLUSIONS
In this paper, a control strategy of the single-phase active
filter using a novel d-q transformation is proposed. The es-
Parameter Value
L
S
1 mH
5.6 %
(at i
sa
= 15 A RMS)
L
AF
1 mH
L
LO
5 mH
C
AF
560 uF
C
LO
3900 uF
AF Switchig Freq F
S
15 kHz
AF Sampring Freq F
C
15 kHz 1 sample delay
LPF Cut-off Freq 35 Hz 2nd order
Compensation Current control
band
800 Hz K
P
=5.0
Compensation Current control
Integral time t
i
100 ms
Phase ditection
Integral time t
1
1 s
AF DC-bus voltag v
DC
*
250 V
Table.1 Parameters of test system.
Fig.9 Waveforms on instantaneous
phase angle detection.
Fig.10 Waveforms on harmonic compensation.

x10rad

rad
e

x30V
1226
order i
La
i
Sa
3 32.31 2.39
5 8.59 1.03
7 3.38 0.36
9 2.29 0.15
11 1.77 0.25
13 1.30 0.63
15 0.99 1.32
17 0.64 1.35
19 0.67 1.03
THD 33.78 3.70
Table.2 Harmonics characteristic.
sential points of this paper are as follows:
1. The proposed method is performed based on Hilbert
Transform and enables to estimate the harmonic com-
ponents both in the steady and transient states.
2. By using the proposed method, the dc bus voltage fluc-
tuation of the active filter can be suppressed, and the
dc bus capacitor can be reduced to 65 % of the conven-
tional size for a given fluctuation specification.
The proposed strategy can be applied to the three-phase
active filter directly, and it can compensate for the harmon-
ics under the unbalanced-three phase condition.
REFERENCES
[1] H.Akagi, Y.Kanazawa and A.Nabae, Instantaneous reactive power
compensators comprising switching devices without energy storage
components, IEEE Trans. Ind.Appl., vol.IA-20, pp.625-631,1984.
[2] H.Akagi, S.Ogasawara and H.Kim, The theory of instantaneous
power in three-phase four-wire systems and its applications, TIEE
Japan, Vol.119-D, No.12,pp.1461-1470, 1999.(In Japanese)
[3] J.L.Willems,A new interpretation of the Akagi-Nabae power com-
ponents for nonsinusoidal three-phase situations, IEEE Trans.
Instrum.Meas., vol.41, no.4, pp.523527-631,1992.
[4] A.Nabae,H.Nakano and S.Togasawa, An instantaneous distortion
current compensator without any coordinate transformation, Pro-
ceedings of IEEJ International Power Electronics Conference-
Yokohama, pp.1651-1655,1995.
[5] T.Tanaka, S.Funabiki and A.Nabae, A new approach to individual-
phase reactive power compensator for non-sinusoidal and unbal-
anced three-phase system, TIEE Japan, Vol.121-D, No.3, pp.308-
390, 2001. (In Japanese)
[6] M.Saitou and T.Shimizu,d-q transformation method for single-
phase power active filter, TIEE Japan, Vol.122-D, No.12, pp.308-
390, 2002. (In Japanese)
[7] M.Saitou and T.Shimzu, A novel control method on single-phase
grid connectable inverter based on Hilbert Transform, Technical
Report of IEICE,Vol.3, pp.308-390, 2001. (In Japanese)
[8] M.Saitou and T.Shimzu, Generalized theory on single-phase ac-
tive and reactive power based on Hilbert Transform, Proceedings
of PESC02,Vol.3, pp.308-390, 2002.
[9] A.V.Oppenheim, R.W.Schafer and J.R.Buck, Discrete-time signal
processing, PRINTICE HALL, 1998.
[10] S.H.Hahn, Hilbert transforms in signal processing, Artech
House,1996.
1227