Rotorside Cascaded PI Controller Design and Gain Tuning for DFIG Wind Turbines
ABSTRACT A novel and efficient solution to the problem of designing and tuning the gains of cascaded PI controllers used in DFIG wind systems is addressed in the present paper. The overall design takes into account the nonlinear dynamic model of the system while the analysis of the innerloop controllers leads to some new concepts, namely that of the effective time constant, which in turn leads to a gain tuning rather independent from the system parameters. The controller structure is completed by exploiting the timescale separation assumption in the design of the outerloop controller. This approach provides the possibility to obtain a simple systematic method for tuning the outerloop controller gains, based on the wellanalyzed in the literature second order transfer function. Hence a more robust control scheme with some essentially enhanced stability and transient properties results, as it is clearly verified by extensive simulation tests.
 Citations (8)
 Cited In (0)

Article: Wind Speed Estimation Based Sensorless Output Maximization Control for a Wind Turbine Driving a DFIG
[Show abstract] [Hide abstract]
ABSTRACT: This paper proposes a wind speed estimation based sensorless maximum wind power tracking control for variablespeed wind turbine generators (WTGs). A specific design of the proposed control algorithm for a wind turbine equipped with a doubly fed induction generator (DFIG) is presented. The aerodynamic characteristics of the wind turbine are approximated by a Gaussian radial basis function network based nonlinear inputoutput mapping. Based on this nonlinear mapping, the wind speed is estimated from the measured generator electrical output power while taking into account the power losses in the WTG and the dynamics of the WTG shaft system. The estimated wind speed is then used to determine the optimal DFIG rotor speed command for maximum wind power extraction. The DFIG speed controller is suitably designed to effectively damp the lowfrequency torsional oscillations. The resulting WTG system delivers maximum electrical power to the grid with high efficiency and high reliability without mechanical anemometers. The validity of the proposed control algorithm is verified by simulation studies on a 3.6MW WTG system. In addition, the effectiveness of the proposed wind speed estimation algorithm is demonstrated by experimental studies on a small emulational WTG system.IEEE Transactions on Power Electronics 06/2008; · 4.08 Impact Factor  SourceAvailable from: Z.Y. DongIEEE Transactions on Power Systems 01/2012; 27(2):713 722. · 2.92 Impact Factor
 Wind Engineering. 01/2004; 28(4):411432.
Page 1
Rotorside Cascaded PI Controller Design and Gain
Tuning for DFIG Wind Turbines
Michael K. Bourdoulis, Student Member, IEEE, and Antonio T. Alexandridis, Member, IEEE
Department of Electrical and Computer Engineering
University of Patras
Rion, Patras, 26500, Greece
{bourdoulis, a.t.alexandridis}@ece.upatras.gr
Abstract—A novel and efficient solution to the problem of
designing and tuning the gains of cascaded PI controllers used in
DFIG wind systems is addressed in the present paper. The
overall design takes into account the nonlinear dynamic model of
the system while the analysis of the innerloop controllers leads to
some new concepts, namely that of the effective time constant,
which in turn leads to a gain tuning rather independent from the
system parameters. The controller structure is completed by
exploiting the timescale separation assumption in the design of
the outerloop controller. This approach provides the possibility
to obtain a simple systematic method for tuning the outerloop
controller gains, based on the wellanalyzed in the literature
second order transfer function. Hence a more robust control
scheme with some essentially enhanced stability and transient
properties results, as it is clearly verified by extensive simulation
tests.
KeywordsDFIG wind turbines; cascaded controllers; PI
controller; controller gain tuning
I.
INTRODUCTION
Climate changes and CO2 emission reduction concerns
have lead electricity grids to take on a distributed generation
structure based on high local renewable energy sources (RES)
penetration. Among the RES connected at the distributed
generation system, variablespeed wind turbines play a key role
due to their size and increased capabilities [1]. Their main task
is to control active and reactive power and especially under
normal operation conditions to provide the optimal or
maximum energy available from the wind. Nowadays, this is
achieved by using large turbines, in the scale of MW, operating
at the variablespeed mode. Variable speed technique is used to
continuously track the generator rotor speed at its optimal value
as it is determined by the real wind speed in a manner that
guarantees the maximum mechanical power extraction [2].
In the frame of variablespeed wind turbine operation,
DoublyFed Induction Generator (DFIG) wind turbines are
widely used, for up to 3MW individual generator ratings
[1],[2]. They provide a possibility of variablespeed regulation
of about ±30% around the synchronous speed. To fulfill the
aim of maximum power tracking, an effective, stable and fast
controller scheme acting on the induction generator rotor side
has to be applied.
Vector or Field Oriented Control (FOC) is the standard
technique used for the control design of induction machines
[2],[3]. In this mode the system frame is aligned with the rotor
or the stator flux (field orientation). Especially for DFIGs, the
standard method used [2][5] is the statorFOC (SFOC). This
prerequires the system model to be represented in the
synchronously rotating dq reference frame. The benefits of this
dq reference model are that both the coupling stator and rotor
coils as well as the steady state variables and inputs (currents,
fluxes and voltages) become constant. As a result, two
controlled inputs are appeared, corresponding to the d and q
components of the rotor voltage, on which usually PI
controllers are implemented. The structure of two cascaded
controllers is used at each input, with the innerloops regulating
the d and q current components while exact cancelation
networks for decoupled operation are utilized [5],[6]. The
dominant controller design method used for the innerloop
controller gains is directly dependent from the generator
parameters. The outerloop PI controllers provide the reference
current values and they are tuned to have a slower response
than those of the innerloop, a feature typical for cascaded
controllers, known as timescale separation assumption. Thus,
the two outerloop PI controllers are used to regulate the
generator’s rotational speed at the optimum (corresponding to
the maximum wind power extraction) and the stator’s reactive
power [4],[5].
Nevertheless, the cascaded controller design is a
challenging problem since the DFIG wind turbine is a
nonlinear system with parameters that may change during the
system operation; there does not exist a systematic and efficient
way of tuning the outer as well as the innerloop gains in a
manner relaxed from the accurate system parameters.
In this paper, an alternative cascaded controller design and
tuning method suitable for DFIG wind turbines is proposed. In
particular, firstly the innerloop controller is substantially
relaxed from the system parameters since the controller design
and gain tuning is based on the concept of the effective time
constant. By introducing this concept, we obtain a controller
with enhanced characteristics which however does not depend
its stability properties on the accurate system parameter values.
Secondly, the outerloop PI controller is slightly modified in a
way that permits its analysis as a secondorder linear system.
In Section II the DFIG model in the dq synchronously
rotating frame is provided. Then, in Section III the DFIG rotor
side control tasks are presented in the SFOC frame. The
proposed control scheme is presented in Section IV, where the
9781467363921/13/$31.00 ©2013 IEEE
POWERENG 2013
733
4th International Conference on Power Engineering, Energy and Electrical Drives
Istanbul, Turkey, 1317 May 2013
Page 2
inner and outerloop controller design and tuning methods are
established. Finally, in Section V the performance of the
closedloop system is evaluated through extensive simulation
results, while in Section VI some conclusions are addressed.
WRIG
Drive Train
Gear Box
Grid
Back?to?back ac/dc
converters with dc?link
and R?L filter
Fig. 1. The DFIG wind system.
II.
DFIG MODELING
A DFIG, as shown in Fig. 1, consists of a wound rotor
induction generator (WRIG) with its stator connected directly
to the grid and its rotor connected to the grid through a
frequency converter, comprising of two backtoback ac/dc
converters with a dclink, and an RL filter at the gridside.
Under balanced threephase ac conditions, the dynamic
model of the DFIG with states and inputs in the dq
synchronously rotating frame can be written as [7]
dsdss qs s ds
R IU
λ
?
ω λ=−+
(a)
qsqss dss qs
R IU
λ
?
ω λ=++
(b)
()
dr drsr qrrdr
VpR I
λ
?
ωω λ=−−+
(c) (1)
()
qr qrsr drr qr
Vp R I
λ
?
ω ω λ=+−+
(d)
errm
TJbT
ω
?
ω=++
(e)
dss ds
L I
L I
m dr
L I
L I
drr drm ds
L I
L I
qss qsm qrqrrqr m qs
L I
L I
λ
λ
λ
λ
=
=
+
+
=
=
+
+
(2)
3
2
()
m
e qs dr
I
λ
ds qr
I
λ
s
L
L
Tp
=−
(3)
3
2
3
2
3
2
3
2
()()
()()
sddsq qssq dsd qs
r dr dr qr qrr qrdr dr qr
PU IU IQ U IU I
P V IV IQV IV I
=+=−
=+=−
(4)
where d and q subscripts stand for the d and qaxis
components, r and s subscripts stand for rotor and stator, U and
V stand for the constant grid voltage and the controlled voltage,
? stands for flux, I stands for current, p stands for the number
of pole pairs, ?r stands for the rotational speed of the generator
rotor, R stands for ohmic resistance, L stands for inductances
and Lm is the mutual inductance, J is the total moment of inertia
at the rotor of the induction generator, b is the damping
coefficient, Te and Tm are the electrical and mechanical torque
at the rotor, while P and Q stand for active and reactive power.
The dot over a symbol stands for the time derivative of the
corresponding variable.
III. DFIG OPERATION IN THE SFOC FRAME AND ROTOR
SIDE CONTROL TASKS
It is wellknown that the DFIG and its controllers is
realized into the standard frame of SFOC technique for the
DFIG wind power systems. This technique continuously
identifies a synchronously rotating flux axis for the DFIG. The
aim of applying SFOC, i.e. accurate identification on the stator
flux axis, is the independent regulation of the electric torque
and the stator flux via the two orthogonal rotor current
components Idr and Iqr, respectively.
To implement the SFOC technique, let the reference frame
be aligned with the daxis component of the stator flux, that
means ?qs=0. This reduces the electromagnetic torque
expression (3) to
3
2
m
eds qr
I
λ
s
L
L
Tp
= −
(5)
It is obvious that for a constant value of ?ds the
electromagnetic torque depends only on the qaxis component
Iqr of the rotor current.
On the other hand, the mechanical power captured from the
wind by the turbine is given by [8]
()
23
wind
1
2
,
mp
PR Cu
ρπλ β
=
(6)
where ? is the air density, R is the radius of the turbine blades,
uwind is the wind speed and Cp is the power coefficient of the
wind turbine, ? is the blade pitch angle and ? is the tipspeed
ratio.
In order to achieve maximum active power extraction, the
maximum value of Cp should be achieved for some constant
?=?opt. This can be achieved through the optimal rotational
speed of the generator, provided by
,
opt
R
g
r optwind
n
u
λ
ω
=
(7)
where ng is the gear ratio due to the existence of a gearbox.
Thus, the wind speed uwind determines through (7) the rotor
generator speed that corresponds to the maximum wind power
extraction. As a result, the rotorside one controller task is to
adapt the generator speed ?r to a reference
,
ref
r r opt
ωω=
.
Taking into account (5), one can see that for any wind
speed uwind the controller should enforce Te to follow Tm by
regulating Iqr to a suitable
qr
I
, as can be realized by an inner
loop PI qaxis current controller. This reference value
obviously comes from an outerloop PI controller which
regulates
r
ω
such that DFIG meets maximum Pm
corresponding to the optimum Tm .
ref
ref
734
4th International Conference on Power Engineering, Energy and Electrical Drives
Istanbul, Turkey, 1317 May 2013
POWERENG 2013
Page 3
In the same way, it can be easily shown that reactive stator
power can be regulated through the Idr innerloop PI controller,
since Qs is dependent from Idr (one can combine (4) with (2) to
reach this relation). A common way to determine
enforcing ?ds to a desired reference value through an outer loop
PI controller. In this paper,
dr
I
is chosen to be constantly zero,
resulting to an almost constant but high value of the stator’s
reactive power.
ref
dr
I
is by
ref
IV. DFIG ROTORSIDE CONTROL SCHEME
Before presenting the proposed control scheme, the system
is transformed to an equivalent and appropriate form for the
analysis. Using (1), (2) and taking into account the reference
frame alignment at the stator flux, i.e.
following dynamic equations for the rotor currents can be
acquired
(
1
r drdrr slip qr
III
στστ ω
+=−
0
qsqs
λ
?
λ
==
, the
)
r
dr
ds
mr
V
RL
σ τ
−
λ
?
+
?
(8)
()
1
qr
r
rqr qrr slip drslip ds
mr
V
R
I
?
II
L
σ τ
−
στστ ω
= −
ωλ
+−+
(9)
where
2
m
1
sr
L L L
σ = −
,
rrr
L R
τ =
and
slipsr
p
ωωω=−
.
From the last expressions it is clear that the dynamics of
and
qr
I are coupled and they can be decoupled by introducing
the following two new control inputs,
dr
I
dr
u and
qr
u , as
()
1
r
dr
drr slip qrds
mr
V
R
uI
L
σ τ
στ ωλ
?
−
=−+
(10)
()
1
qr
r
qrr slip drslip ds
mr
V
R
uI
L
σ τ
στ ω = −ωλ
−
−+
(11)
Through (10) and (11), (8) and (9) are simplified to
r drdr dr
I
?
Iu
στ+=
(12)
r qrqr qr
I
?
Iu
στ+=
(13)
which represent two decoupled and identical first order
equations.
As shown in [6],
its derivative has a negligible average value. This leads to the
realization that the control input expressions can be simplified
to
ds
λ has a fairly constant value, ˆ
ds
λ , while
dr
drr slip qr
r
V
R
uI
στ ω=+
(14)
()
1
ˆ
qr
r
qrr slip drslip ds
mr
V
R
uI
L
σ τ
στ ω= −ωλ
−
−+
(15)
A. Rotorside controllers’ structure
Due to the fact that equations (12) and (13) for
dynamics are decoupled first order differential equations, the
following PI current controllers for
dr
I and
qr
I
dr
u and
qr
u are proposed
0
()()
t
ref
dr
ref
dr drPdr dr Idrdr
ukIIkIIdτ=−+−
?
(16)
0
()()
t
ref
qr
ref
qrqrPqrqrIqrqr
ukIIkIIdτ=−+−
?
(17)
where
representing the proportional and integrator gains, respectively.
Pdr
k ,
Pqr
k and
Idr
k,
Iqr
k are positive scalars
The daxis current reference,
zero [6]. On the other hand, the qaxis current reference,
is provided by a slightly modified outerloop PI controller as
follows
ref
dr
I, is chosen to be equal to
ref
qr
I ,
0(
?
)
t
ref
qr
ref
rPrIr
Ikkd
ωω
ωωωτ = −+−
(18)
where
proportional and integrator gain, respectively, while
generator’s rotor rotational speed reference provided by (7).
P kω and
Ikω are positive scalars representing the
ref
r
ω
is the
B. Innerloop Controller Tuning
It is obvious from (12) and (13) that the dynamics of the
rotor currents are identical to each other when the new control
inputs given by (10) and (11) are applied with innerloop
controllers provided by (16) and (17). Thus the d and q
subscripts are omitted for remainder of the tuning analysis
providing the following expressions of the current dynamics
and the innerloop controllers
rrrr
I
?
Iu
στ+=
(19)
0
()()
t
ref
r
ref
rr PrrIrr
ukIIkII dτ=−+−
?
(20)
The Laplace transformations of (19) and (20) can be easily
calculated and manipulated to arrive at
( )
s
2
(1)
r PrIr
ref
rr Pr Ir
I k s
+
k
Isksk
στ
+
=
++
which can be equivalently written in the following form
( )
s
1
( )1
r
ref
rir
I
H ss
I
τ
=
+
(21)
where
arbitrary time constant resulting from the arbitrary selected
value of
Ir
k , while ( )H s represents a transfer function given
by
ir
τ
is calculated as 1
irIr
k
τ =
and represents an
1
2
1
1
( )
T s
T s
H s
+
+
=
(22)
with
735
4th International Conference on Power Engineering, Energy and Electrical Drives
Istanbul, Turkey, 1317 May 2013
POWERENG 2013
Page 4
1r
T
στ=
and
2
Pr
Ir
k
k
T
=
(23)
Transfer function (22) can represent a lead or lag
compensator if
12
TT
>
or
12
TT
<
unitary if
12
TT
=
and constitute the conventional design [6].
From the definition of
as
Pr
k becomes larger the transfer function
lag compensator. Obviously, for small values of
may represent a lead compensator.
, respectively, or even it can be
1T and
2T it is clear that, for a given
H s determines a
Ir
k ,
( )
Pr
k , ( )H s
As one can see from (21),
multiplier  amplifier  on the time constant
the effective time constant,
( )
H s acts as a dynamic
τ [9]. Defining
ir
eff
ir
τ
, as
( )
eff
ir ir
H s
ττ=
(24)
a lead operation of
frequencies, while at low frequencies
hand, a lag operation of
frequencies while, again, at low frequencies
shown in Fig. 2. A lag operation of
in increased damping of transients since it provides an effective
time constant
ir
τ
smaller at high frequencies, where transients
occur. On the other hand, a lead operation of
to result in poorly damped transients since it provides an
effective time constant
ir
τ
greater at high frequencies.
Therefore, lag operation is adopted since it is expected to
enhance transient performance.
( )
H s increases the value of
eff
ir
τ
at high
eff
irir
ττ=
; on the other
τ
at high
ττ=
( )
H s reduces the value of
eff
ir
eff
ir ir
, as
( )
H s is expected to result
eff
( )
H s is expected
eff
( )H ω
1
1
2
Τ
Τ
2
1
Τ
1
1
Τ
ω
Fig. 2. Asymptotic Bode plot for magnitude absolute value (lag operation).
Hence, for lag operation of ( )
H s ,
2T should be determined
as
2rr
Ta στ=
with 1
ra >
Taking into account (23), the arbitrary selection of
related with
Ir
k , σ and
Pr
k may be
rτ as follows
PrrrIrrIr
kakk
στστ=>
(25)
Remark 1. If
transfer function with controller gains dependent from the
exact system parameters σ and
coincides with the standard technique used [6]. The proposed
design allows a small integral gain
manner that ensures reduced sensitivity at steadystate and the
flexibility to select
Pr
k large enough corresponding to some
arbitrary
1
r
α > . This approach permits a rather independent
controller design from the accurate values of the system
parameters. Simultaneously, since the time constant
clearly decreased when transients occur, a faster response and
decay than that of the initially selected
1
r
α = then
eff
irir
ττ=
, (21) reduces to a first order
rτ , and the proposed design
Ir
k to be selected in a
eff
ir
τ
is
ir
τ is expected.
C. Outerloop Controller Tuning
Since the SFOC technique used results in a stator flux ˆ
which is fairly constant, the electromagnetic torque expression
(5) can be written as
ds
λ
3
2
ˆ
m
e ds qr
I
λ
s
L
L
Tp
= −
and is directly dependent from the qaxis rotor current. Tuning
the innerloop current controller to be faster enough than the
outerloop, then one can consider
separate timescale assumption. Thus, the simplified block
diagram of Fig. 3 is valid.
ref
qrqr
II
=
by using the
ref
qrqr
II
≡
+−
+−
eT
m
T
1
+
Jsb
r
ω
ref
r
ω
3
2
ˆ
m
ds
s
L
pL
λ
+−
Ik
s
ω
P kω
Fig. 3. Block diagram of the subsystem and its controller.
Considering the mechanical torque,
the following transfer function is acquired for the closedloop
system
m
T , as a disturbance,
2
3
2
L
ˆ
3
2
3
2
ˆˆ
m
dsI
sr
ref
r
mm
dsPdsI
ss
L
p
J L
k
L
b
J
p
J L
p
J L
sksk
ω
ωω
λ
ω
ω
λλ
=
?
?
?
?
?
?
+++
(26)
It is obvious that the last transfer function’s form allows the
control designer to reach the following desired closedloop
transfer function with ζ damping factor,
and unity static gain
n
ω natural frequency
2
n
22
n
2
r
ref
rn
ss
ω
ζω
ω
ω
ω
=
++
(27)
where
2
n
3
2
ˆ
m
dsI
s
L
p
J L
k
ω
ωλ
=
and
3
2
ˆ
2
m
n dsP
s
L
b
J
p
J L
k
ω
ζωλ
=+
.
736
4th International Conference on Power Engineering, Energy and Electrical Drives
Istanbul, Turkey, 1317 May 2013
POWERENG 2013
Page 5
Applying for example the 2% criterion [10], the desired
settling time
sT is connected to ζ and
n
ω through
4
sn
T
ζω=
.
Also, for the desired damping factor
loop controller gains are calculated as
des
ζ
choice, the outer
8
T
2
3
1
ˆ
ds
λ
s
P
sm
L
b
J
J
p L
k
ω
?
?
?
?
?
?
=−
(28)
22
des
2
3
1
ˆ
ds
λ
16
ζ
s
I
m
s
L
J
p L
k
T
ω
=
(29)
Remark 2. Usually, the outerloop controller gains are selected
either empirically or by using complicated methods such as the
Symmetric Optimum Method [11]. In our case a simple gain
tuning method is proposed based on a flexible criterion used in
a wellknown linear system technique [10]. Again, a design
lightly dependent from the system parameters is achieved.
V.
SIMULATION RESULTS
In this section, the effectiveness of the proposed design and
analysis is verified through extensive simulation results. The
response of a 2MW DFIG is simulated for the case of step
changes in the wind speed, as shown in Fig. 4. All the
parameters of the simulated system are provided in Table I. For
100
the innerloop proportional gain is selected to be
10
IdrIqr
kk
==
and through (25)
outerloop controller gains are calculated for
ζ=
and ˆ
3.17
ds
λ =
through (28) and (29) to be equal
to
87.84
and
70.3
. The innerloop low frequency
time constant is equal to 0.1sec while the high frequency
effective time constant is equal to 0.001sec. Two cases are
examined: i) with input including simplified decoupling as
given by (14), (15) and ii) without taking into account
decoupling, i.e.
drr dr
V R u
=
and
ra =
1.033
PdrPqr
kk
==
, while the
5sec
sT =
,
0.707
des
P kω=
Ikω=
qrr qr
V R u
=
.
TABLE I. 2MW DFIG WIND SYSTEM PARAMETERS
Symbol Quantity Value
s R
stator resistance 0.01 ?
r R
rotor resistance 0.00842 ?
sL
stator inductance 0.005305 H
rL
L
p
R
rotor inductance
0.0053137 H
m
mutual inductance
pole pairs
blade radius
rotor friction
coefficient
rotor inertia
0.0051839 H
3
35 m
b
0.00015 N*m*sec/rad
J
765.6 kg*m2
g n
β
ρ
λ
gearbox ratio 62.5
0o pitch angle
air density 1.2 kg/m3
6.325
opt
optimal tipspeedratio
s
ω
grid angular frequency 2?50 rad/sec
2
ds
2
qs
UU
+
grid voltage 700 V (rms)
05 1015 2025 3035 404550
8
9
10
11
12
time (sec)
uwind (m /sec)
Fig. 4. Wind speed profile.
0
51015 2025 30 35 4045 50
90
95
100
105
110
115
120
125
130
time (sec)
?
r, ?
r r e f ( r a d / s e c )
?r with decoupling
?r without decoupling
?rref
Fig. 5. DFIG rotor angular speed vs. its reference.
05101520 253035 40 4550
3.16
3.17
3.18
3.19
?ds (Wb)
05101520253035404550
4
2
0
2x 10
3
time (sec)
?qs (Wb)
with decoupling
without decoupling
Fig. 6. Stator flux component responses.
05 101520253035404550
50
0
50
100
150
Idr (A)
05101520253035404550
0
500
1000
1500
time (sec)
Iqr (A)
with decoupling
without decoupling
Fig. 7. Rotor current component responses.
The response of the rotor rotation speed versus its reference
is provided in Fig. 5. Figures 6 and 7 provide the responses of
the d and qaxis stator flux and rotor current components,
737
4th International Conference on Power Engineering, Energy and Electrical Drives
Istanbul, Turkey, 1317 May 2013
POWERENG 2013
Page 6
respectively. It is clear that the cascaded controller design as
provided in Section IV is fully confirmed for the case where
decoupling networks are utilized. The daxis flux component is
approximately constant at the ˆ
λ value selected, while the q
axis flux component is effectively regulated to zero. The flux
and current responses appear to have a fast and smooth
transient for the case (i), while for case (ii) the system
robustness permits convergence to the steadystate equilibrium
after some oscillations with a slower transient.
ds
The stator and rotor active and reactive power responses are
provided in Fig. 8 and 9, respectively. As expected, the rotor
reactive power is minimized while the stator reactive power
experiences a large but almost constant at steady state value.
Finally, in Fig. 10 a zoom on the transient response of both
rotor currents is presented for different values of the
parameter, for case (i). It becomes clear that, as expected, when
α increases the transient response is enhanced with more
smooth and smaller in magnitude transients, while the settling
time is unaffected.
r
α
r
0510 15 20253035 404550
0
0.5
1
1.5
2x 10
6
Ps (W)
05 1015 2025 3035 404550
1
0
1
2
3x 10
5
time (sec)
Pr (W)
with decoupling
without decoupling
Fig. 8. Stator and rotor active power responses.
05 1015 2025 303540 4550
10
9
8
7x 10
5
Qs (Var)
05 10152025 30 3540 4550
2
0
2
4x 10
4
time (sec)
Qr (Var)
with decoupling
without decoupling
Fig. 9. Stator and rotor reactive power responses.
9.51010.5 1111.5 1212.513 13.5
100
0
100
200
Idr (A)
9.51010.5 1111.5 1212.5 1313.5
600
800
1000
1200
time (sec)
Iqr (A)
?r=1
?r=25
?r=50
?r=100
Fig. 10. Transient response detail of rotor current components.
VI. CONCLUSIONS
A more flexible design and innovative tuning method for
the conventional cascaded rotorside control system of a DFIG
is proposed. Through extensive simulation results, enhanced
and robust transient performance has been demonstrated even
in the case where decoupling innerloop networks are omitted.
REFERENCES
[1] R. Teodorescu, M. Liserre, and P. Rodriguez, Grid Converters for
Photovoltaic and Wind Power Systems. West Sussex, UK: Wiley
IEEE Press, 2011.
[2] G. Abad, G. Iwanski, J. López, L. Marroyo, and M.A. Rodríguez,
Doubly Fed Induction Machine: Modeling and Control for Wind
Energy Generation Applications. Hoboken, NJ: Wiley, 2011.
[3] R. Pena, J.C. Clare, and G.M. Asher, “Doubly fed induction generator
using backtoback PWM converters and its application to variable
speed windenergy generation,” Proc. IEE Elect. Power Appl., vol.
143, no. 3, pp. 231–241, May 1996.
[4] A.D. Hansen, P. Sørensen, F. Iov, and F. Blaabjerg, “Control of
variable speed wind turbines with doublyfed induction generators,”
Wind Eng., vol. 28, no. 4, pp. 411–434, June 2004.
[5] W. Qiao, W. Zhou, J.M. Aller, and R.G. Harley, “Wind Speed
Estimation Based Sensorless Output Maximization Control for a Wind
Turbine Driving a DFIG,” IEEE Trans. Power Electron., vol. 23, no.
3, pp. 1156–1169, May 2008.
[6] A. Yazdani, and R. Iravani, VoltageSourced Converters: Modeling,
Control, and Applications. Hoboken, NJ: Wiley, 2010.
[7] P.C. Krause, O. Wasynczuk, and S.D. Sudhoff, Analysis of Electric
Machinery and Drive Systems, 2nd Edition. Piscataway, NJ: Wiley
IEEE Press, 2002.
[8] L. Yang, Z. Xu, J. Østergaard, Z.Y. Dong, and K.P. Wong, “Advanced
Control Strategy of DFIG Wind Turbines for Power System Fault
Ride Through,” IEEE Trans. Power Systems, vol. 27, no. 2, pp. 713–
722, May 2012.
[9] I.C. Konstantakopoulos, M.K. Bourdoulis, and A.T. Alexandridis,
“An Alternative PI Controller Design Approach for PWM regulated
ac/dc threephase Converters,” IEEE Intern. Conf. on Industrial Tech.
(ICIT 2012), pp. 955960, Athens, Greece, Mar. 1921, 2012.
[10] J.J. D’Azzo, and C.H. Houpis, Linear Control System Analysis and
Design: Conventional and Modern, 2nd Edition. McGrawHill, 1981.
[11] J.W. Umland, and M. Safiuddin, “Magnitude and symmetric optimum
criterion for the design of linear control systems: what is it and how
does it compare with the others?,” IEEE Trans. Ind. Appl., vol.26, no.
3, pp. 489 497, May/Jun 1990.
738
4th International Conference on Power Engineering, Energy and Electrical Drives
Istanbul, Turkey, 1317 May 2013
POWERENG 2013