Sensorless Closed-Loop Voltage and Frequency Control of Stand-Alone Dfigs Introducing Direct Flux-Vector Control
Sensorless Closed-Loop Voltage and Frequency Control of Stand-Alone Dfigs Introducing Direct Flux-Vector Control
Sensorless Closed-Loop Voltage and Frequency Control of Stand-Alone Dfigs Introducing Direct Flux-Vector Control
fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2019.2955421, IEEE
Transactions on Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS
Wind DFIG
Stator
Abstract— An innovative control method, termed direct
is Loads
flux-vector control (DFC), is proposed for the stand-alone
Cf
doubly-fed induction generators (DFIGs) feeding local AC- GEARBOX
loads. In these systems, both the amplitude and frequency RSC Vdc LSC
of the stator-voltage must be precisely controlled. The ir
Rotor
proposed method, instead of controlling the rotor currents
or voltages, directly controls the rotor flux-vector,
including magnitude and angle. To achieve an accurate DSP controllers
control, two separate closed-loop hysteresis controllers Fig. 1. Generic conceptual diagram of the stand-alone DFIG system.
are employed in which the stator-voltage amplitude and converter approximately 20-30% of the total generator output
frequency are adjusted through the rotor-flux magnitude
power, wide control capabilities and high efficiency [2].
and angle, respectively. The proposed control method is
performed in the rotor reference frame, thus it does not Over the last decade, research community have mainly
require the rotor speed/position sensors or any reference- focused on the grid-connected DFIG applications [3]-[7],
frame transformation. As an outstanding feature, the where the main purpose is control of active and reactive
method works for both sub- and super-synchronous powers. However, the generator should be able to feed local
speed modes without changing the switching table. loads during the grid-faults [8]. Moreover, there are a large
Hence, no need to detect the operation mode of DFIG. number of inherent islanded-loads such as remote villages,
Besides, the method consists of a simple implementation electric ships and military equipment, which require the stand-
and is almost parameter-independent, as it only requires alone operation of DFIGs. To fulfill these requirements and
the rotor resistance. The proposed method is implemented
exploit DFIGs full capacity, efforts are directed toward the
on a 3-kW laboratory scale DFIG and its performance,
effectiveness, robustness and correctness are evaluated stand-alone operation mode of DFIG [9]-[27]. Unlike the grid-
and discussed for a series of experimental tests. connected mode, the main goal for the stand-alone operation is
to generate the constant stator-voltage amplitude and
Index Terms— Doubly-fed induction generators, direct frequency, under variable conditions such as wind-speed
control, sensorless voltage control, stand-alone operation, changes and load variations [9]. Therefore, control strategies
voltage and frequency control, wind energy conversion. are completely different in stand-alone operation.
A generic system of the stand-alone DFIG applications is
I. INTRODUCTION shown in Fig. 1, where the stator feeds local loads. The rotor
winding is fed through a back-to-back converter including
W IND energy conversion systems (WECSs) have become
very attractive due to the growing concerns regarding
air pollution, advances in power electronics and low
load side converter (LSC) and rotor side converter (RSC),
controlled via digital signal processor (DSP). The LSC is
generation costs [1]. In the existing wind energy applications, connected to the stator via a transformer, which is responsible
the doubly-fed induction generators (DFIGs) have been widely for the DC-link voltage regulation, regardless of power flow
used, because of their advantages in terms of fractionally-sized direction in RSC [9]. Also, it can be adopted as an active filter
to compensate current-harmonics under nonlinear loads [10].
To control stator-voltage amplitude and frequency, proper
voltage waveforms are injected into the rotor windings
Manuscript received Jan 02, 2019; revised Apr 21, 2019 and Sep through the RSC [2]. Capacitor filters are paralleled to the
28, 2019; accepted Oct 27, 2019.
Majid Rostami is with the Department of Electrical Engineering,
stator, to compensate part of the excitation current and remove
Faculty of Engineering, University of Isfahan, Iran, (e-mail: switching-frequency harmonics caused by LSC and RSC [11].
majid.rostami@eng.ui.ac.ir, majdrostami2008@gmail.com). Stand-alone DFIGs come in two types namely DFIG-DC
Sul Ademi is with the Warwick Manufacturing Group, University of and DFIG-AC [12]. In DFIG-DC type, the stator is connected
Warwick, Coventry, U.K (e-mail: s.ademi@warwick.ac.uk).
Seyed M. Madani is with the Department of Electrical Engineering,
to DC loads through a rectifier. In this type, DC bus voltage
Faculty of Engineering, University of Isfahan, Iran, (corresponding magnitude must be controlled accurately, whereas, the voltage
author, email: m.madani@eng.ui.ac.ir, madani104@yahoo.com).
0278-0046 (c) 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2019.2955421, IEEE
Transactions on Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS
frequency only needs to be maintained around the rated direct rotor-flux control has not been considered so far for
frequency to extract the highest power from generator [13]. stand-alone DFIG-ACs.
Several field-oriented control (FOC) based schemes are The proposed method consists of two hysteresis controllers
reported for DFIG-DC, where some schemes aimed to reduce which are employed to regulate the magnitude and angle of
voltage sensors [14] or eliminate speed sensor [15], [16]. A the rotor-flux. The stator-voltage amplitude is adjusted by the
direct torque control (DTC) scheme is also presented in [13] rotor-flux magnitude, while the stator-voltage frequency is
which has different switching tables for sub- and super- regulated by the rotor-flux angle. The most important and
synchronous operation modes and uses a mechanical speed- novel features of the proposed DFC method are as follows:
sensor to determine the operation mode of DFIG. On the other - It does not need the rotor position and speed. Therefore, no
hand, some researches have recently published on stand-alone mechanical speed-sensor or speed-estimator is required.
DFIG-AC types, where the stator is connected to the local AC- - It is robust against the system uncertainties since both
loads. Unlike the DFIG-DC, in DFIG-AC systems, in addition voltage amplitude and frequency have closed-loop controllers
to the stator-voltage amplitude, the frequency must be with real-time feedbacks.
precisely adjusted, with less than 1% error [17]. - It is almost parameter-independent, as it only needs the
The basic topologies and control methods for stand-alone rotor resistance.
DFIG-ACs are presented in [18]. Some researches have aimed - It does not require any reference-frame transformation.
to mitigate harmonics of stator currents and voltages, under - Unlike [13], its switching strategy is the same for both the
nonlinear and unbalanced loads [19]-[22]. A resonant sub- and super-synchronous speed operation modes. As a
controller for RSC is presented in [19], to current harmonic result, there is no need to determine the DFIG operation-mode.
elimination under nonlinear loads. This method is expanded in - The stator frequency is estimated by voltage space-vector,
[20] by considering the unbalanced loads. Several other therefore no need to high-computational PLL algorithms.
techniques are also employed for feeding unbalanced loads This paper is organized as follows. Section II describes the
such as repetitive control [21] and predictive control [22]. equivalent circuit model of the stand-alone DFIG-AC systems.
The aforementioned literature are based on FOC-method, Different parts of the proposed controller including regulator
which requires the rotor-position acquired from shaft-mounted parts and estimator parts are presented in section III and
mechanical sensors. Elimination of such sensors brings many section IV, respectively. The performance of the proposed
advantages in terms of cost, maintenance, sturdiness and method are verified by a series of experimental test. Moreover,
cabling requirements. The latter deficiency may be the proposed method is experimentally compared with the
particularly severe with DFIG where regular brush servicing FOC-based method presented in [20]. The results are reported
can pose a growing risk of sensor failure. For this reason, and discussed in section V. Section VI draws conclusions.
several methods are presented for software-estimation of the
rotor-position [23]-[25]. Yet, such methods use the state- II. SYSTEM MODEL DESCRIPTION AND ANALYSIS
equations of system, which are highly parameter-dependent The general equivalent circuit model of stand-alone DFIG-
[26]. Thus, the rotor-position estimation is negatively affected AC system in arbitrary reference frame by rotation speed of 𝜔
by the parameters changes during DFIG operation, which is depicted in Fig. 2.
highly reduces the reliability of the controller. This model consists of two loops that represent the rotor
As another major drawback of FOC-based methods, the
stator-voltage frequency is controlled by getting feedback ir rr j(ω-ω r ) λ r L lr L ls jω λ s r s
is
from the mechanical rotor-speed, which acts as an open-loop
controller, thus the speed-sensor errors directly affects the
control loop. Due to the system uncertainties, including vr p λr Lm p λs vs ZL
communication noises, rotor speed-sensor errors and sudden Rotor-loop Stator-loop
wind-speed changes, the open-loop controller is not accurate
enough. To solve this, an autonomous frequency control is Fig. 2. Equivalent circuit of DFIG in arbitrary reference frame by
suggested in [27], which gets feedback directly from the rotation speed of 𝜔.
stator-voltage frequency. Although this method acts as a and stator windings which are linked together through the
closed-loop controller, it suffers from the rotor position magnetizing inductance 𝐿𝑚 . The voltage equations of the
sensor, high computational phased looked loop (PLL) stator and rotor loops are as,
algorithm, and five proportional integral (PI) controllers. 𝑣⃗𝑠 = 𝑟𝑠 𝑖̇⃗𝑠 + 𝑗𝜔𝜆⃗𝑠 + 𝑝𝜆⃗𝑠 (1)
Consequently, this paper, attempts to fill the void by
⃗ ⃗
𝑣⃗𝑟 = 𝑟𝑟 𝑖̇⃗𝑟 − 𝑗(𝜔−𝜔𝑟 )𝜆𝑟 + 𝑝𝜆𝑟 (2)
proposing a new and comprehensive control method named
direct flux-vector control (DFC) for RSC of stand-alone where 𝑣⃗𝑠 and 𝑣⃗𝑟 are the stator and rotor voltage-vectors, and
DFIG-AC systems. The proposed controller is sensorless and 𝜆⃗𝑠 and 𝜆⃗𝑟 are the stator and rotor fluxe-vectors which are
has separate closed-loop regulators for stator-voltage expressed as,
frequency as well as stator-voltage amplitude, which 𝜆⃗𝑠 = 𝐿𝑚 𝑖̇⃗𝑟 + 𝐿𝑠 𝑖̇⃗𝑠 (3)
eliminates both drawbacks of the FOC-based methods
simultaneously. As another contribution of this paper, instead 𝜆⃗𝑟 = 𝐿𝑟 𝑖̇⃗𝑟 + 𝐿𝑚 𝑖̇⃗𝑠 (4)
of the rotor voltages or currents, the proposed method directly where 𝐿𝑠 and 𝐿𝑟 are the stator and rotor inductances and are
controls the rotor-flux vector. To the author’s best knowledge, defined as,
0278-0046 (c) 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2019.2955421, IEEE
Transactions on Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS
𝐿𝑠 = 𝐿𝑙𝑠 + 𝐿𝑚 (5) stator-voltage 𝑣⃗𝑠 is calculated versus the stator-flux 𝜆⃗𝑠 . The
𝐿𝑟 = 𝐿𝑙𝑟 + 𝐿𝑚 (6) equivalent circuit of Fig. 2 is transferred to the stator-flux
reference frame, by choosing 𝜔 = 𝜔𝑒 , thus under steady-state,
where 𝐿𝑙𝑠 and 𝐿𝑙𝑟 are the stator and rotor leakage inductances.
the stator-flux becomes constant (dc value), meaning that,
By choosing the appropriate reference-frame speed 𝜔,
voltage source of either rotor or stator loop can be eliminated, 𝑝𝜆⃗𝑠 = 0 (7)
which simplifies the relationship between the system On the other hand, considering the load 𝑍𝐿 , the stator
variables. In the stationary reference frame (𝜔 = 0), the current 𝑖̇⃗𝑠 can be obtained versus the stator voltage 𝑣⃗𝑠 as,
voltage source of the stator-loop is eliminated, while the 𝑣⃗𝑠
voltage source of rotor-loop is eliminated in the rotor 𝑖̇⃗𝑠 = − (8)
𝑍𝐿
reference frame (𝜔 = 𝜔𝑟 ). Rearranging the voltage equation of stator loop (1) yields,
The power flow in DFIG depends on the rotor-speed. In the 𝑝𝜆⃗𝑠 = −𝑗𝜔𝑒 𝜆⃗𝑠 − 𝑟𝑠 𝑖̇⃗𝑠 + 𝑣⃗𝑠 (9)
sub-synchronous speeds, the load is supplied only from the By substituting (7) and (8) in (9), the relation between the
stator, whereas, in super-synchronous speeds the load is stator-voltage and stator-flux becomes,
supplied from both the stator and rotor. 1 𝑟𝑠 1
𝜆⃗𝑠 = (1 + )𝑣⃗𝑠 = 𝑘 𝑣⃗ (10)
𝑗𝜔𝑒 𝑍𝐿 𝑗𝜔𝑒 𝐿 𝑠
III. PROPOSED REGULATOR PARTS OF THE CONTROLLER
where, 𝑘𝐿 is defined as,
A. Overview 𝑟𝑠
𝑘𝐿 ≜ (1 + ) (11)
The proposed controller design is depicted in Fig. 3, 𝑍𝐿
whereby the stator-voltage amplitude and frequency are To find the relationship between the stator-flux 𝜆⃗𝑠 and
regulated by two separated closed-loops. The upper loop is rotor-flux 𝜆⃗𝑟 , the midpart of the circuit in Fig. 2, is re-drawn
responsible for adjusting the voltage amplitude by controlling as illustrated in Fig. 4(a). As such, the derivative operator 𝑝
the rotor-flux magnitude, where a larger rotor-flux generates a can be removed from all the voltage sources and inductances.
larger stator-voltage amplitude. While the frequency is tuned This converts the inductances to resistances as shown in Fig.
in the lower loop by controlling the rotor-flux angle. Each of 4(b). Furthermore, an appropriate Thevenin equivalent results
these loops contains a PI controller and a hysteresis-controller. in the equivalent circuit of Fig. 4(c), which clearly depicts the
The output signals of the hysteresis-controllers (𝑞𝜆 and 𝑞𝜃 ) relationship between the rotor-flux and stator-flux.
determine if the rotor-flux magnitude |𝜆⃗𝑟 | and angle 𝜃𝜆𝑟 i r Llr Lls i s i r Llr Lls i s σ Lr i r
should be decreased or increased. Based on these signals and
rotor-flux position, the appropriate voltage-vector is selected Lm λ
from the switching-table and is applied to rotor, by inverter.
pλr Lm pλ s λr Lm λs λr s
Ls
The proposed control method consists of two kind of parts:
regulator parts and estimator parts, which are specified in Fig. (a) Thevenin (b) (c)
3. The regulator parts are detailed in the following sub- Fig. 4. Midpart of the equivalent circuit of stand-alone DFIG.
sections as being the main parts of the proposed controller, From the same figure it can be noted that 𝜎 is the leakage
while the estimator parts are explained in section-IV. factor and is expressed as,
B. Proposed stator-voltage amplitude regulator 𝐿2𝑚
𝜎 =1− (12)
In the proposed method, the stator-voltage amplitude is 𝐿𝑠 𝐿𝑟
regulated by adjusting the rotor-flux magnitude, as shown in According to Fig. 4(c), stator-flux and rotor-flux
the upper control-loop of Fig. 3. The purpose of this sub- relationship becomes,
section is to illustrate the relationship between the stator- 𝐿𝑠
𝜆⃗𝑠 = (𝜆⃗ − 𝜎𝐿𝑟 𝑖̇⃗𝑟 ) (13)
voltage amplitude and the rotor-flux magnitude. Initially the 𝐿𝑚 𝑟
0278-0046 (c) 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2019.2955421, IEEE
Transactions on Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS
By substituting (10) in (13), 𝑣⃗𝑠 can be obtained versus the the stator-voltage frequency, where its output is the reference
∗ ∗
rotor-flux 𝜆⃗𝑟 as, of slip-frequency 𝜔𝑠𝑙 . As shown in Fig. 3, by integrating 𝜔𝑠𝑙 ,
∗
𝐿𝑠 𝜔𝑒 the reference of rotor-flux angle 𝜃𝜆𝑟 is obtained,
𝑣⃗𝑠 = 𝑗 (𝜆⃗ − 𝜎𝐿𝑟 𝑖̇⃗𝑟 ) (14)
𝑘𝐿 𝐿𝑚 𝑟 ∗
𝜃𝜆𝑟 ∗
= ∫ 𝜔𝑠𝑙 𝑑𝑡 (24)
Substituting 𝜆⃗𝑟 from (4) in (14) yields, Based on (23) and (24), the small-signal model of the
𝐿𝑠 𝜔𝑒
𝑣⃗𝑠 = 𝑗 [(1 − 𝜎)𝐿𝑟 𝑖̇⃗𝑟 + 𝐿𝑚 𝑖̇⃗𝑠 ] (15) frequency regulator is depicted in Fig. 6.
𝑘𝐿 𝐿𝑚 PI
Assuming 𝜎 ≪ 1 and using (4), (15) can be approximated: Δω*s Δωsl* 1 Δϴλr* Δω sl Δωs
Cω = k pω + kiω s
𝐿𝑠 𝜔𝑒 𝐿𝑠 𝜔𝑒
𝑣⃗𝑠 ≅ 𝑗 [𝐿𝑟 𝑖̇⃗𝑟 + 𝐿𝑚 𝑖̇⃗𝑠 ] = 𝑗 𝜆⃗ (16) s s
𝑘𝐿 𝐿𝑚 𝑘𝐿 𝐿𝑚 𝑟 Δωs Δωm
Therefore,
Fig. 6. Small-signal model of the proposed voltage frequency regulator.
𝐿𝑠 𝜔𝑒
|𝑣⃗𝑠 | = |𝜆⃗ | (17) According to Fig. 6, the relationship between the ∆𝜔𝑠 and
|𝑘𝐿 |𝐿𝑚 𝑟
Equation (17) shows that stator-voltage amplitude |𝑣⃗𝑠 | has its reference value ∆𝜔𝑠∗ is,
𝐶𝜔 1
direct relationship with rotor-flux magnitude |𝜆⃗𝑟 |, therefore, a ∆𝜔𝑠 = ∆𝜔𝑠∗ + ∆𝜔𝑚 (25)
PI controller can be employed to regulate |𝑣⃗𝑠 | by controlling 1 + 𝐶𝜔 1 + 𝐶𝜔
𝑘
|𝜆⃗𝑟 |. To investigate performance of PI controller, the small- By substituting 𝐶𝜔 = (𝑘𝑝𝜔 + 𝑖𝜔 ) in (25),
𝑠
signal model for |𝑣⃗𝑠 | can be obtained by linearizing (17) as, 𝑘𝑝𝜔 𝑠 + 𝑘𝑖𝜔 ∗
𝐿𝑠 ∆𝜔𝑠 = ∆𝜔𝑠
(1 + 𝑘𝑝𝜔 )𝑠 + 𝑘𝑖𝜔
∆|𝑣⃗𝑠 | = (𝜔 ∆|𝜆⃗𝑟 | + |𝜆⃗𝑟 |∆𝜔𝑒 ) (18) 𝑠 (26)
|𝑘𝐿 |𝐿𝑚 𝑒 + ∆𝜔𝑚
Since 𝜔𝑒 is a controlled variable, its variation ∆𝜔𝑒 is (1 + 𝑘𝑝𝜔 )𝑠 + 𝑘𝑖𝜔
negligible and (18) can be written as, The steady-state gain of the frequency regulator can be
𝐿𝑠 𝜔𝑒 obtained in (26) by replacing the Laplace variable with zero,
∆|𝑣⃗𝑠 | = ∆|𝜆⃗𝑟 | (19) ∆𝜔𝑠
|𝑘𝐿 |𝐿𝑚 (𝑠 → 0) = 1 (27)
Based on (19), the small-signal model of the stator-voltage ∆𝜔𝑠∗
amplitude regulator is depicted in Fig. 5. This confirms the appropriate performance of the frequency
PI regulator in steady-state.
Δ |Vs |* Δ|λr |* Δ|Vs |
Cv = k pv + kiv Pv = Ls ω e D. Proposed hysteresis design and switching strategy
s |kL |Lm To regulate the rotor-flux magnitude and angle, a two-level
Δ |Vs | hysteresis-controller is used in each control-loop. The
Fig. 5. Small-signal model of the proposed voltage amplitude regulator. magnitude and angle comparators are depicted in Fig. 7(a) and
Considering Fig. 5, the relationship between ∆|𝑣⃗𝑠 | and its 7(b), respectively.
qλ λ |λr |*
reference value ∆|𝑣⃗𝑠 |∗ is, 1 1
-1 1 |λr |
∆|𝑣⃗𝑠 |
1 -1
𝐶𝑣 𝑃𝑣
=
-1
(20) Δλ
1 -1 1
∆|𝑣⃗𝑠 | ∗ 1 + 𝐶𝑣 𝑃𝑣 0 λb t
𝑘 -λb λb 0 -1
1
Substituting 𝐶𝑣 = (𝑘𝑝𝑣 + 𝑖𝑣) in (20) yields,
-1
𝑠 -1 -1 1 1
1
∆|𝑣⃗𝑠 | 0 Δϴ 1
0
0 t
(𝑠 → 0) = 1 (22) -ϴb ϴb 0
∆|𝑣⃗𝑠 |∗ (b)
This shows that the steady-state error of the regulator is zero. Fig. 7. Design of the hysteresis-comparators. (a) rotor-flux magnitude
In the other words, the stator-voltage amplitude |𝑣⃗𝑠 | follows comparator and (b) rotor-flux angle comparator.
its reference value |𝑣⃗𝑠 |∗ . The output of rotor-flux magnitude hysteresis controller is,
C. Proposed stator-voltage frequency regulator 1 𝑖𝑓 Δ𝜆𝑟 = |𝜆⃗𝑟 |∗ − |𝜆⃗𝑟 | > 𝜆𝑏
𝑞𝜆 = { (28)
In the proposed method, the stator-voltage frequency 𝜔𝑠 is −1 𝑖𝑓 Δ𝜆𝑟 = |𝜆⃗𝑟 |∗ − |𝜆⃗𝑟 | ≤ (−𝜆𝑏 )
regulated by the rotor-flux angle. This frequency is the sum of and the rotor-flux angle hysteresis controller output is,
the mechanical rotor-speed 𝜔𝑚 and the slip frequency 𝜔𝑠𝑙 , ∗
1 𝑖𝑓 ∆𝜃 = 𝜃𝜆𝑟 − 𝜃𝜆𝑟 > 𝜃𝑏
𝜔𝑠 = 𝜔𝑚 + 𝜔𝑠𝑙 (23) 𝑞𝜃 = { ∗ (29)
0 𝑖𝑓 ∆𝜃 = 𝜃𝜆𝑟 − 𝜃𝜆𝑟 ≤ (−𝜃𝑏 )
According to (23), to keep 𝜔𝑠 constant, 𝜔𝑚 changes must
be compensated by controlling 𝜔𝑠𝑙 . Since 𝜔𝑠 has direct where 𝜆𝑏 and 𝜃𝑏 are the permitted deviation (band values) of
relationship with 𝜔𝑠𝑙 , a PI controller can be used to regulate the rotor-flux magnitude and angle, respectively.
0278-0046 (c) 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2019.2955421, IEEE
Transactions on Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS
Switching states of rotor side inverter is displayed in Fig. A. Stator-voltage amplitude and frequency estimation
8(a). When each of the rotor phases are connected to the 𝑉𝑑𝑐 To avoid time-consuming PLL algorithms, the proposed
or ground, the corresponding switch is labeled 1 or zero, method estimates the stator-voltage frequency, using the
respectively. Based on the switching states, 8 voltage-vectors voltage space-vector, which can be determined by,
can be applied on the rotor windings; six non-zero (𝑉1 − 𝑉6 ) 𝑣⃗𝑠 = |𝑣⃗𝑠 |∠𝜃𝑠 = 𝑣𝑎𝑠 + 𝛼𝑣𝑏𝑠 + 𝛼 2 𝑣𝑐𝑠 (30)
and two zero vectors (𝑉0 , 𝑉7 ), as depicted in Fig. 8(b). The 2𝜋
phasor-plane is divided into six 60-degree sectors, S1 to S6, where, 𝛼 = 1∠ , |𝑣⃗𝑠 | and 𝜃𝑠 are the stator-voltage amplitude
3
where the voltage-vector 𝑉𝑖 is located at the middle of sector and angle in the stator reference frame. The stator voltages
Si. At each sampling instant, the rotor-flux vector angle 𝑣𝑎𝑏𝑐𝑠 are obtained through the voltage sensors. The stator
determines the present sector Si. Depending on the number of frequency can be obtained by derivation of 𝜃𝑠 as,
this sector and the comparators outputs, the appropriate 𝑑𝜃𝑠
𝜔𝑠 = (31)
voltage-vector is selected and is applied to the rotor windings. 𝑑𝑡
V3(010) V2(110) As seen in Fig. 3, a low-pass filter (LPF) is used to
+ eliminate the switching frequency noises.
B S3 S2
A u1 1 u2 1 u3 1
Vdc V4(011) S4 S1 V1(100) B. Rotor-flux magnitude and angle estimation
Rotor 0 0 0
C windings
_ S5 S6 By considering the equivalent circuit model of Fig. 2 in
V5(001) V6(101) rotor reference frame (by substituting 𝜔 = 𝜔𝑟 ), rotor fluxes
(a) (b) can be obtained from the voltage-equation of the rotor-loop as,
Fig. 8. (a) Switching states and (b) phasor-plane and voltage vectors.
To investigate the effect of each voltage-vector, assume the 𝜆𝑎𝑏𝑐𝑟 = ∫(𝑣𝑎𝑏𝑐𝑟 − 𝑟𝑟 𝑖𝑎𝑏𝑐𝑟 )𝑑𝑡 (32)
rotor-flux is in sector S1. Applying 𝑉2 or 𝑉6 to the rotor where the rotor currents 𝑖𝑎𝑏𝑐𝑟 are measured by sensors, while
windings increases the rotor-flux magnitude, whereas, the rotor voltages 𝑣𝑎𝑏𝑐𝑟 are calculated as described in the next
applying 𝑉3 or 𝑉5 decreases the rotor-flux magnitude. By sub-section. As mentioned earlier, the rotor-flux space vector
controlling the rotor-flux magnitude, the stator-voltage can be obtained as,
amplitude can be adjusted as explained in the sub-section B.
𝜆⃗𝑟 = |𝜆⃗𝑟 |∠𝜃𝜆𝑟 = 𝜆𝑎𝑟 + 𝛼𝜆𝑏𝑟 + 𝛼 2 𝜆𝑐𝑟 (33)
On the other hand, applying 𝑉2 or 𝑉3 increases the rotor-flux
angle, whereas, applying 𝑉5 or 𝑉6 decreases the rotor-flux where |𝜆⃗𝑟 | and 𝜃𝜆𝑟 are the rotor-flux magnitude and angle in
angle. By controlling the rotor-flux angle, the stator-voltage the rotor reference frame.
frequency can be adjusted as explained in the sub-section C. In C. Rotor-voltage estimation
the same way, the appropriate voltage-vectors that need to be
applied to the rotor according to the stator-voltage amplitude Since the rotor voltages contain large switching ripples,
and frequency conditions can also be obtained for the other instead of using voltage sensors, the rotor voltages are
sectors, as shown in Table-I. calculated based on the DC-bus voltage 𝑉𝑑𝑐 and switching
states, using the following expressions.
TABLE I
𝑉𝑑𝑐
SWITCHING STRATEGY OF THE PROPOSED CONTROL METHOD 𝑣𝑎𝑟 = (2𝑢1 − 𝑢2 − 𝑢3 ) (34)
𝑞𝜆 𝑞𝜃 S1 S2 S3 S4 S5 S6 3
𝑉𝑑𝑐
1 1 𝑉2 𝑉3 𝑉4 𝑉5 𝑉6 𝑉1 𝑣𝑏𝑟 = (−𝑢1 + 2𝑢2 − 𝑢3 ) (35)
3
𝑉𝑑𝑐
1 0 𝑉6 𝑉1 𝑉2 𝑉3 𝑉4 𝑉5 𝑣𝑐𝑟 = (−𝑢1 − 𝑢2 + 2𝑢3 ) (36)
3
-1 1 𝑉3 𝑉4 𝑉5 𝑉6 𝑉1 𝑉2 where, a switching state uj is equal to one/zero if the phase j
-1 0 𝑉5 𝑉6 𝑉1 𝑉2 𝑉3 𝑉4 is connected to the 𝑉𝑑𝑐 /ground, respectively.
It is worth highlighting that unlike [13] approach, the V. EXPERIMENTAL VERIFICATION
proposed method works for both sub- and super-synchronous
operation modes, without changing the switching table and as A. Setup description
a result, it is not required to detect the DFIG operation mode. The schematic block diagram of the test-setup used for
experimental validation is shown in Fig. 9.
IV. ESTIMATOR PARTS OF THE CONTROLLER Rectifier
Prime mover
RSC
dc-link
To accurately control of the stator-voltage amplitude and Variac
frequency, closed-loop controllers are employed. Therefore, DFIG AC
the real-time feedback of the stator-voltage amplitude and Loads
Independent
frequency are required to be compared with their reference source Dissipating
iar ibr Neutral vas vbs
values. Moreover, as presented in the previous section, the real resistors
time values of the rotor-flux magnitude and angle are required Gate Drive Analog Signal Conditioning
for the regulator parts (see Fig. 3). For this purpose, the u1..3
proposed method employs the most effective and light TMS320F28335 DSP
parameter-dependent existing estimation-methods which are RSC Controller (see Fig. 3)
briefly explained in the following sub-sections. Fig. 9. Schematic block diagram of the DFIG experimental setup.
0278-0046 (c) 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2019.2955421, IEEE
Transactions on Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS
Since this paper focus is on controlling the rotor-side B. Wide mechanical rotor-speed change around the
converter, without affecting the overall system performance, synchronous speed
some parts of the commercial DFIGs of Fig. 1 have been Figure 11, illustrates the performance of the proposed
replaced in the test-setup with similar alternatives as follows: method during the wind speed change around the synchronous
- The load-side converter is replaced by a rectifier supplied speed 1000 rpm. For this purpose, the wind-speed is
from an independent source, as per [21]-[22]. considered to be changed such that the corresponding rotor-
- A set of resistors as a dissipating resistive-load is installed speed increases from 900 to 1100 rpm, while the voltage
in parallel with the dc-link capacitor. In super-synchronous reference and load are kept constant. It can be realized that by
mode, the surplus power is injected to the dc-link through the increasing the wind-speed, frequency of the injected rotor-
RSC, and then will be dumped in the dissipating resistors. currents are decreased until the synchronous speed which are
This enables the system to operate in super-synchronous dc values. By passing through the synchronous speed, the
mode, without involving in the control of load-side converter. phase sequence of the rotor currents is changed. The sector-
- An induction machine controlled by a separate commercial number, which represents the rotor-flux position at each time
drive is employed as a wind-turbine emulator. ascends in sub-synchronous speed operation, whereas, it is
As depicted in Fig. 9, the data collected from the sensors are changed to descending in super-synchronous speed operation.
initially converted to a suitable voltage level for DSP by an This means that the direction of rotor-flux vector is reversed
analogue signal conditioning circuit. Control signals from when the rotor-speed exceeds the synchronous-speed.
DSP are applied to RSC through a gate drive circuit, which
Moreover, the rotor-flux magnitude |𝜆⃗𝑟 | follows its reference
acts as a buffer and isolator.
within a band-margin which is determined by Δ𝜆𝑟 in the
Figure 10, shows the 3-kW laboratory scaled DFIG
hysteresis-controller. It is worth mentioning that rotor-flux
prototype used for experimental validation purposes, which its
magnitude decreases when the wind-speed increases, which
parameters are enlisted in the Appendix. Three 200-W light
prevents overvoltage in the stator windings.
bulbs are connected to the stator as local AC-loads. Moreover,
Figure 11, also depicts the stator-voltage amplitude and
the stator is equipped with a 3-phase, Y-connected, 150 𝜇𝐹
frequency waveforms. By increasing wind-speed, at first, the
filtering capacitor. Rotor-currents and stator-voltages are
stator-voltage amplitude and frequency are slightly increased.
measured by LA55-p and LV25-p LEM transducers,
Then, they are returned back to their reference values, in
respectively. The proposed control method is implemented on
steady-state. This test confirms the robustness of the proposed
a TMS320F28335 DSP with a sampling frequency of 10 kHz.
control method during wind-speed changes in both sub- and
The control signals are applied to a six-pulse inverter
super-synchronous speeds.
connected to the rotor. A Data Acquisition (DAQ) card made
by AdvantechTM is used to gather the data from voltage and
current sensors. To verify the effectiveness of the proposed
Rotor currents (2A/div)
1.02
Dissipating Voltage
Rotor flux (pu) Sector number Rotor speed Frequency (pu) Voltage
0.98
1.02
DFIG 1
Prime
Mover 0.98
Capacitors
Commerical
1100
(rpm)
Light Bulbs
drive 1000
900
Fig. 10. DFIG test setup used for experimental validation purposes.
5
The rotor-flux estimation, based on (32), is the only block 3
which has parameter dependency (see Fig. 3). Therefore, the 1
only parameter uncertainty that can degrade the controller 0.98
performance is the rotor-resistance. To investigate the
0.94
parameter-dependency of the proposed method, a sensitivity
analysis is performed on rotor-resistance in sub-section E, as 0.9
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
per [28]. Finally, in sub-section F, a comparison is made with Time (s)
the FOC-based method presented in [20]. Fig. 11. Experimental results for mechanical rotor-speed change test.
0278-0046 (c) 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2019.2955421, IEEE
Transactions on Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS
amplitude (pu)
amplitude (pu)
1.02 1.02
Sector number Rotor speed Frequency (pu) Voltage
Sector number Rotor speed Frequency (pu) Voltage
1 1
0.98 0.98
0.96 0.96
1.02 1.02
1 1
0.98 0.98
1100 1100
(rpm)
(rpm)
1000 1000
900 900
800 800
5 5
3 3
1 1
0.98 0.98
dc-link voltage Rotor flux
dc-link voltage Rotor flux
(pu)
(pu)
0.94 0.94
0.9 0.9
0.98 0.98
(pu)
(pu)
0.94 0.94
0.9 0.9
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time (s) Time (s)
(a) (b)
Fig. 12. Experimental results for load-change test. (a) sub-synchronous operation mode and (b) super-synchronous operation mode.
0278-0046 (c) 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2019.2955421, IEEE
Transactions on Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS
±20% of its real value and the results are depicted in Fig. 15.
1.03
0.97
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
amplitude (pu)
0.95
Since non-linear loads are not the aim of this paper, without
Reference
0.9 Voltage
affecting the overall performance of this method, those parts
0.85 are not considered here for comparison. The comparison
Sector number Rotor flux (pu)
1
results, including the rotor-currents and stator-voltage
0.9 amplitude and frequency, are shown in Fig. 16.
0.8
According to Fig. 16, in both control methods, increasing
the load causes the rotor-currents to increase. However,
5
3 Proposed method
FOC-based method
1
Rotor currents (2A/div)
1.1
illustrates the effectiveness of the proposed method for control
Voltage
0278-0046 (c) 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2019.2955421, IEEE
Transactions on Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS
compared to the FOC method, the proposed controller has directly controls the rotor flux-vector. This method controls
faster reaction/response, which is an inherent characteristic of the stator-voltage amplitude and frequency by controlling the
DTC-based methods. Because of this, the stator-voltage rotor-flux magnitude and angle, respectively. Since the
amplitude and frequency waveforms of the proposed method proposed method is completely performed in the rotor
have smaller undershoot and are returned back faster to their reference-frame, neither rotor position/speed sensor nor any
reference values. Fig. 16, illustrates that the proposed method reference-frame transformation is required. The proposed
has a better performance even during the stage when the new- method employs the same switching strategy for both the sub-
load has been removed. and super-synchronous speed operation-modes, which is
Figure 17, compares the parameter-dependency among the another outstanding feature of this controller design.
proposed controller and that of FOC-based method. For this Therefore, no operation-mode detection is required. A series
purpose, in the load-change test, all of the DFIG parameter of experimental results have demonstrated the high
values in both controllers are increased +20% of their real performance, and effectiveness of the proposed control
values, and the related results are evidenced. It has been found method on a 3-kW proof-of-concept DFIG prototype. As
that compared to the FOC-based method, the proposed method being largely parameter independent, the controller has shown
is more robust against parameter uncertainty. to be robust, versatile, and applicable to a variety of small-to-
Proposed method FOC-based method large-scale generators and/or drive systems.
amplitude (pu)
1.1
Voltage
1
0.9 Reference
APPENDIX
0.8
Voltage
0.7 DFIG Prototype Characteristics (Parameters and Ratings)
Frequency (pu)
0278-0046 (c) 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2019.2955421, IEEE
Transactions on Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS
[10] M. Pattnaik and D. Kastha, "Harmonic compensation with zero- Power Delivery, vol. 28, DOI 10.1109/TPWRD.2013.2243170, no. 2,
sequence load voltage control in a speed-sensorless DFIG-based stand- pp. 752-760, April 2013.
alone VSCF generating system," in IEEE Transactions on Industrial [28] R. Sadeghi, S. M. Madani, M. Ataei, M. R. Agha Kashkooli and S.
Electronics, vol. 60, DOI 10.1109/TIE.2012.2235396, no. 12, pp. 5506- Ademi, "Super-Twisting Sliding Mode Direct Power Control of a
5514, Dec. 2013. Brushless Doubly Fed Induction Generator," in IEEE Transactions on
[11] G. Iwanski and W. Koczara, "Sensorless direct voltage control of the Industrial Electronics, vol. 65, DOI 10.1109/TIE.2018.2818672, no. 11,
stand-alone slip-ring induction generator," in IEEE Transactions on pp. 9147-9156, Nov. 2018.
Industrial Electronics, vol. 54, DOI 10.1109/TIE.2007.892746, no. 2,
pp. 1237-1239, April 2007. Majid Rostami was born in Najafabad, Iran, in
[12] M. F. Iacchetti and G. D. Marques, "Voltage control in a DFIG-DC 1990.
system connected to a stand-alone dc load," in 9th International He received the B.Sc. degree from University
Conference on Compatibility and Power Electronics (CPE), Costa da of Isfahan, Isfahan, Iran, in 2012, the M.Sc.
Caparica, DOI 10.1109/CPE.2015.7231094, pp. 323-328, 2015. degree from Isfahan University of Technology,
[13] A. Gundavarapu, H. Misra and A. K. Jain, "Direct torque control scheme Isfahan, Iran, in 2014, both in Electrical Power
for DC voltage regulation of the standalone DFIG-DC system," in IEEE Engineering. Currently, he is a Ph.D. student in
Transactions on Industrial Electronics, vol. 64, DOI Electrical Power Engineering at University of
10.1109/TIE.2016.2644623, no. 5, pp. 3502-3512, May 2017. Isfahan, Isfahan, Iran.
[14] H. Misra and A. K. Jain, "Analysis of stand-alone DFIG-DC system and His research interests are power electronics
DC voltage regulation with reduced sensors," in IEEE Transactions on and electrical machine drives especially DFIGs and PMs.
Industrial Electronics, vol. 64, DOI 10.1109/TIE.2017.2669889, no. 6,
pp. 4402-4412, June 2017.
[15] C. Wu and H. Nian, "Sinusoidal current operation of a DFIG-DC system Seyed M. Madani received the B.Sc. degree
without stator voltage sensors," in IEEE Transactions on Industrial from Sharif University of Technology, Tehran, in
Electronics, vol. 65, DOI 10.1109/TIE.2017.2786259, no. 8, pp. 6250- 1989, the M.Sc. degree from University of
6258, Aug. 2018. Tehran, Tehran, in 1991, and the Ph.D. degree
[16] G. D. Marques and M. F. Iacchetti, "Sensorless frequency and voltage from Eindhoven University of Technology, the
control in the stand-alone DFIG-DC system," in IEEE Transactions on Netherlands, in 1999, all in electrical power
Industrial Electronics, vol. 64, DOI 10.1109/TIE.2016.2624262, no. 3, engineering.
pp. 1949-1957, March 2017. From 2000 to 2005, he worked in Texas AM
[17] K. Vijayakumar, N. Kumaresan and N. Ammasai Gounden, "Operation University, University of Puerto Rico and
and closed-loop control of wind-driven stand-alone doubly fed induction University of Wisconsin at Madison, as assistant
generators using a single inverter-battery system," in IET Electric Power professor or visiting professor. From 2005 to 2011, he worked at
Applications, vol. 6, DOI 10.1049/iet-epa.2011.0204, no. 3, pp. 162-171, Isfahan University of Technology as Assistant Professor. He is
March 2012. currently an Associate Professor at University of Isfahan, Isfahan.
[18] R. D. Shukla and R. K. Tripathi, "Topologies for stand-alone DFIG His research interests include renewable energy, electrical
based wind energy conversion system," in 2nd International Conference machines and drive.
on Power, Control and Embedded Systems, Allahabad, DOI 10.1049/iet-
epa.2011.0204, pp. 1-5, 2012.
[19] V. Phan and H. Lee, "Control strategy for harmonic elimination in stand- Sul Ademi received the B.Eng. and Ph.D.
alone DFIG applications with nonlinear loads," in IEEE Transactions on degrees in electrical and electronics
Power Electronics, vol. 26, DOI 10.1109/TPEL.2011.2123921, no. 9, engineering from Northumbria University,
pp. 2662-2675, Sept. 2011. Newcastle upon Tyne, U.K., in 2011 and 2014,
[20] V. Phan and H. Lee, "Performance enhancement of stand-alone DFIG respectively.
systems with control of rotor and load side converters using resonant From 2015 to 2017, he was a Lead
controllers," in IEEE Transactions on Industry Applications, vol. 48, Researcher, engaged in knowledge exchange
DOI 10.1109/TIA.2011.2175883, no. 1, pp. 199-210, Jan.-Feb. 2012. between University of Strathclyde, Glasgow,
[21] F. Wei, X. Zhang, D. M. Vilathgamuwa, S. S. Choi and S. Wang, U.K. and GE Grid Solutions, Stafford, U.K. He is
"Mitigation of distorted and unbalanced stator voltage of stand-alone currently a Research Scientist with the Warwick
doubly fed induction generators using repetitive control technique," in Manufacturing Group (WMG), University of Warwick, Coventry, U.K.
IET Electric Power Applications, vol. 7, DOI 10.1049/iet- His current research interest include electric drives, control of
epa.2012.0317, no. 8, pp. 654-663, September 2013. doubly fed machines, fault tolerant electric power steering applications,
[22] V. Phan and H. Lee, "Improved predictive current control for and design/analysis of novel permanent-magnet machines.
unbalanced stand-alone doubly-fed induction generator-based wind
power systems," in IET Electric Power Applications, vol. 5, DOI
10.1049/iet-epa.2010.0107, no. 3, pp. 275-287, March 2011.
[23] A. B. Ataji, Y. Miura, T. Ise and H. Tanaka, "Direct voltage control with
slip angle estimation to extend the range of supported asymmetric loads
for stand-alone DFIG," in IEEE Transactions on Power Electronics, vol.
31, DOI 10.1109/TPEL.2015.2414481, no. 2, pp. 1015-1025, Feb. 2016.
[24] P. K. Gayena, D. Chatterjeeb and S. K. Goswamib, "An improved rotor
position estimator for the control of DFIG based on stator EMF phasor,"
2015 Annual IEEE India Conference (INDICON), New Delhi, DOI
10.1109/INDICON.2015.7443480, pp. 1-6, 2015.
[25] M. K. Malakar, P. Tripathy and S. Krishnaswamy, "A predictor-
corrector based rotor slip-position estimation technique for a DFIG,"
2017 7th International Conference on Power Systems (ICPS), Pune, DOI
10.1109/ICPES.2017.8387331, pp. 424-429, 2017.
[26] A. Parida and D. Chatterjee, "A robust parameter non-sensitive rotor
position and speed estimator for DFIG," Proceedings of the 2014
International Conference on Control, Instrumentation, Energy and
Communication (CIEC), Calcutta, DOI 10.1109/CIEC.2014.6959040,
pp. 12-16, 2014.
[27] Y. Zhang and B. T. Ooi, "Stand-alone doubly-fed induction generators
(DFIGs) with autonomous frequency control," in IEEE Transactions on
0278-0046 (c) 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.
Alternative Proxies: