International Journal of Advanced Research in Engineering and Technology (IJARET)

Volume 8, Issue 4, Jul - Aug 2017, pp. 95–110, Article ID: IJARET_08_04_009
Available online at http://iaeme.com/Home/issue/IJARET?Volume=8&Issue=4
ISSN Print: 0976-6480 and ISSN Online: 0976-6499
© IAEME Publication

APPLICATION OF EKF BASED DYNAMIC

STATE ESTIMATION FOR DFIG ROTOR
POWER CONTROL UNDER FAULTY CURRENT
SENSOR MEASUREMENTS
Dishang D. Trivedi, Urmil B. Bhatt and Santosh C. Vora
Department of Electrical Engineering,
Institute of Technology, Nirma University, Ahmedabad
Gujarat, India

ABSTRACT
Ever increasing demand of energy and limited availability of fossil fuel have
encouraged significant penetration of renewable energy sources. The doubly fed
induction generator (DFIG) based wind power has emerged as one of the substantial
alternative among renewable energy sources, due to controllability of electrical power
over wide wind speed range. Control circuits of DFIG emphatically rely on the
feedback from voltage, current measurements and the DFIG performance is likely to
be affected by the sensor failure, sensor dynamic characteristics, and their interface
circuits. If current measurement issue / sensor faults are not detected timely, then it can
lead to generator disconnection from grid. To overcome unforeseen shutdown and loss
of revenue, it is imperative to have sensor fault detection and back-up arrangement.
Present work investigates effect of rotor current sensor failure on DFIG operation and
proposes use of dynamic state estimation (DSE) to overcome adversarial effect of faulty
current sensor. Dynamic states of DFIG are estimated using extended Kalman filter

currents (   and quadrature axis stator and rotor currents (    ) of DFIG are
(EKF) algorithm. Non-measurable dynamic states viz. direct axis stator and rotor

estimated using active power (P) and reactive power (as measurement input to EKF
algorithm. Use of dynamically estimated states as a back-up to measurements of rotor
control circuit is proposed under the conditions of faulty current sensors like noisy
measurement output, measurement with outliers or measurement unavailability to
provide continuous control. Simulation results of proposed approach is presented on
DFIG based single-machine-infinite-bus (SMIB) system using MATLAB/Simulink
platform.
Key word: Doubly Fed Induction Generator, Extended Kalman Filter, Dynamic State
Estimation, Current Sensor, Measurement Noise.

http://iaeme.com/Home/journal/IJARET 95 [email protected]
 Application of Ekf Based Dynamic State Estimation For Dfig Rotor Power Control Under Faulty
Current Sensor Measurements

Cite this Article: Dishang D. Trivedi, Urmil B. Bhatt and Santosh C. Vora, Application
of EKF Based Dynamic State Estimation For DFIG Rotor Power Control Under Faulty
Current Sensor Measurements. International Journal of Advanced Research in
Engineering and Technology, 8(4), 2017, pp 95–110.
http://iaeme.com/Home/issue/IJARET?Volume=8&Issue=4

   Stator and rotor voltage

NOMENCLATURE

   Stator and rotor current

   Stator and rotor flux linkages
   Stator and rotor resistance
   Synchronous speed and rotor speed
    Stator, rotor and magnetizing inductance
   Stator active and reactive power
   Rotor active and reactive power
  Total output active and reactive power

    = Estimated  using EKF algorithm

    = Estimated  using EKF algorithm
   = Rotor active and reactive power obtained using   

 Synchronous reference frame coordinate – direct and quadrature axis

SUBSCRIPTS

 Stator and rotor quantities

 Instance of sample in discrete time systems

Priori
SUPERSCRIPTS


Posteriori
 Estimated signal
Reference signals

1. INTRODUCTION
Increasing penetration of wind power to conventional power system has raised concerns
regarding aspects of power system e.g. system frequency and inertia control, reactive power
contributions, system stability etc. [1]. Among available wind generator technologies, doubly
fed induction generator (DFIG) is one of the preferred wind power generators in the booming
renewable power scenario [2]. It is important to understand nonlinear behavior of a machine
and hence a mathematical model and / or the state space model of DFIG is essential.
Comparison of performance of 3 rd order and 5th order mathematical model of DFIG under
various dynamic conditions is presented in [3]. Mathematical models of DFIG with different
control strategies are presented in [4-7]. Eigenvalue sensitivity analysis and its impact on fault
ride through capability of DFIG can be found in [8 - 9]. DFIG operation is mainly controlled
by two back-to-back converters. Solution to fault in internal circuit of DFIG i.e. fault in back-
to-back converter is well presented using Fourier supported analysis in [10]. Performance of
DFIG largely depends on condition of network to which it is connected, as well as on faithful

http://iaeme.com/Home/journal/IJARET 96 [email protected]
 Dishang D. Trivedi, Urmil B. Bhatt and Santosh C. Vora

operation of its own internal elements viz. back-to-back converters, voltage, current and speed
measurement devices. Solution to faulty speed measurement using fuzzy based sliding mode
observer (FSMO) control of DFIG is proposed in [11]. FSMO approach estimates the speed of
DFIG for control and eliminates requirement of speed sensor [11]. After detection of fault in
sensors, reconfiguration of closed loop control using bank of observers is proposed in [12].
Bunch of observers generate residual signal after fault detection and reconfiguration using
estimated variables by observers is proposed in [12]. To overcome effect of fault(s) in current
sensors, hardware approach with field programmable gate array (FPGA) based grid side
converter control is well proposed in [13]. Using state space based model of DFIG, current
sensor coherent observer is proposed in [14] using MATLAB/Simulink platform. Robust
observer presents substitute to current sensor under dynamic and transient condition, but differs
from actual quantity from current sensors under dynamic condition [14].
Increasing presence of DFIG based wind power has made it essential for EMS operator to
have knowledge of dynamic states of DFIG along with conventional synchronous generators.
Reason for increasing avenues of research in dynamic state estimation (DSE) for DFIG states
is its highly nonlinear model, involving complex equivalent circuit and existence of converters

estimation of      and  (rotor speed) using extended Kalman filter (EKF) over sub-
along with rotor side as well as grid side controllers [15]. El-hagry and Eskander presented

synchronous to super-synchronous speed range of DFIG while employing ,  and capacitor

charging current  as measurement variables [16]. Further, estimated       and  of
DFIG are used to derive      and ! . These derived variables are consequently used

DFIG [16]. Estimation of five states of DFIG viz.       ,  and a parameter J (moment
in the control loop to obtain control parameters to regulate rotor voltage and rotor position of

of inertia)using ensemble Kalman filter (EnKF) is well documented in [17]. Fan et. al. gave
insight to parameter calibration along with estimation of parameter. Sensitivity of EnKF
algorithm to different measurement noise levels, initial state and parameter errors are discussed

β, DSE of stator and rotor current "# and"$ , rotor speed  and its position, % with augmented
in [17]. With the help of state space model of DFIG with reference to stationary axis viz. α and

integrals of "# and "$ is conceptualized using EKF for SMIB system in [18].
With the sudden variations in active and reactive powers supplied to point of common
coupling (PCC), control circuit of DFIG undergoes rapid variations of different control
variables. Quick and frequent changes in control variables is also supported by fast operation
of sensors. This may cause mal-operation or failure of sensors. For better stability and control
of DFIG, use of accurately estimated dynamic states, in case of a mal-functioned current sensor
conditions, offers an opportunity of exploration. The noisy measurement case, measurements
with outliers and / or condition of measurement data unavailability is simulated for faulty
current sensor operations. Output of EKF based estimation algorithm is proposed as substitute
of mal-functioned current sensor measurements for controlling of DFIG rotor circuit behaviour.
In brief, application of estimated output using EKF algorithm for better control of DFIG based
wind power forms core of presented work.
The rest of the paper is organized as follows. A mathematical approach for DFIG model
is presented in Section II. Description of simulation preliminaries, discrete mathematical model
of DFIG and EKF implementation for DSE of DFIG forms Section III. Section IV is case
studies imbibed with results and discussion followed by a conclusion in Section V.

http://iaeme.com/Home/journal/IJARET 97 [email protected]
 Application of Ekf Based Dynamic State Estimation For Dfig Rotor Power Control Under Faulty
Current Sensor Measurements

2. MATHEMATICAL APPROACH FOR DSE OF DFIG MODEL

A DFIG system connected to an infinite bus is considered as shown in Fig.1. Stator terminals
of the induction generator are connected to the grid at the point of common coupling (through
a converter transformer) and a slip-ring rotor through a back-to-back converter.

Figure 1 DFIG connected to infinite bus (Source: Internet).

Generally, DFIG operating speed is restricted to ±30 % of synchronous speed. During wide
variation in speed, two main components viz. rotor side converter (RSC) and Grid side
converter (GSC) plays significant role in injecting power at grid frequency. In the present

explained through block diagram representation in Fig. 2. The          

&'(
research, rotor side control circuit is presented for analysis. Operation of RSC control circuit is

represent pre-determined and/or derived reference signals.

Figure 2 Rotor side converter (RSC) control circuit for DPC

assumed to be   )!SFO allows decoupled control of  and  by employing measurement

Considering stator field oriented (SFO) DFIG operation, the stator direct axis voltage is

signals  and  respectively as shown in Fig. 2.

Total simulation duration considered is 6s for all the cases. Relatively small simulation
period supports considering wind speed to be constant during simulation. The dynamics and
mathematical modelling of mechanical components of system is not been focused. Currents
entering in to machine is positive i.e. motor convention are adopted for the DFIG operation as
denoted in Fig.1. As shown in control circuit (Fig. 2), outer measurement loop continuously
monitors  is observed continuously by inner

 and whereas for RSC operation rotor current 

http://iaeme.com/Home/journal/IJARET 98 [email protected]
 Dishang D. Trivedi, Urmil B. Bhatt and Santosh C. Vora

measurement loop using current sensor. The rotor current measurements   and 
are fed to
the RSC control circuit to control active and reactive power output. Primary focus of presented
work is application of resulting (estimated) states as EKF output for controlling of rotor power.
The linearized operation of controllers and control circuit is considered as the parameter
variations are slower as compared to controller actions.
The voltage equations for the DFIG as an induction machine are available in [17], and are
presented here (Eqn. (1 – 4)).

    *   +

,-.

(1)

     +   +
,/.

(2)

    * 0 *   +

, -1

(3)

    * 0 *   +

, /1

Here,       are the flux linkages,       are the stator and
(4)

rotor voltages on synchronously rotating d and q axis. The  and  are the speed of
synchronous frame and rotor respectively.
Stator and rotor flux linkages are given by,
     +        +    (5)

    +         +   (6)

Rotor current components viz. direct and quadrature axis components, are the vital variables
to control active and reactive power output of DFIG [6]. These current are considered here as
state variables along with direct and quadrature axis components of stator current. Differential
equation representing stator and rotor current variables and state matrix are given by Eqn. (7),

8 <
7 9 ; α=*   + 2 > +  0  * > 5 +    +     +    *   ?
7 ;  
4 8 <
; 8 <
7  4 7 α=* 2 > +  0   * > 5  *    *     +    +    *   ? ;
234 5  7 9 ;  7 α=   *     *    + 2 0  * > *   5 *   +   ? ;
7 ;  
7

; 7 4 ;
7                      
;
7 9 ; 64 :    +    *2  > +  0  * > 5 *    *   + 
6  ? :
7 ;
α=                   

6 9 :
(7)

1
α =
L s Lr − L 2
m
where,

Active and reactive power output of DFIG is used as measurement variable for EKF based
dynamic state estimation. Stator active and reactive power of DFIG is presented as,

  0  +   

@
>
(8)

A
  =   + ?
B

http://iaeme.com/Home/journal/IJARET 99 [email protected]
 Application of Ekf Based Dynamic State Estimation For Dfig Rotor Power Control Under Faulty
Current Sensor Measurements

Rotor active and reactive power is given by,

  > 0   +  
@
(9)

A
  0 *    
B
Total active and reactive power output form DFIG is formed by following equations,

   +   C=   +  + 0  +  ?D

@
>
(10)

A
   +    00  *   + 0  *   
B
Measurement matrix 2E5 for EKF algorithm is,
 @ =  +  +   +    ?
2E5  F G  H   I
 > 0  *   +   *     
(11)

Stator and rotor voltages referred to and axis are input to EKF based DSE algorithm.
Hence, input vector J is given by,


2K5  L  M


(12)

3. EKF IMPLEMENTATION FOR DFIG

Power system formulations consists of nonlinear relationships between the state and the
measurements or it may happen that the process itself is nonlinear in nature. Hence, it is
required to imbibe these nonlinearities in estimation process to have accurate estimation of
power system state variables [19]. In presented work, EKF – a variant of Kalman filter
approach is used for dynamic state estimation of DFIG based wind generating system
connected to infinite bus (as SMIB). The estimated results are then used to improve power
output under noisy current measurement, bad measurement and measurement unavailable
conditions. DFIG based power system modelling and necessary mathematical steps for EKF
implementation are covered in literature [20-21]. A detailed description of mathematical steps
involved in EKF implementation is presented in Appendix A for reader’s reference.

3.1. Discrete model of DFIG for EKF implementation

Stator currents    and rotor currents    are the states to be estimated using DSE
To perform EKF base state estimation algorithm, discretized model of DFIG is employed.

algorithm. Active power of DFIG 0 and reactive power 0 are the measurement variables.
Differential algebraic equations (DAEs) presented in previous section are discretized for DSE.
State equations shown in Eqn. (7) and measurement equations in Eqn. (10) are represented in
discretized form respectively as Eqn. (13) and (14 – 15),
 
8 N< 8 OPQ <
7 O ;= 7  OPQ ; + R=
7 ; 7  ; NS   OPQ   OPQ  OPQ  J NS ?! T9
7 O ; 7 OPQ ;
(13)

6 O : 6  OPQ :

N   > 0O O + O O +  O   O +  O  N 
@
(14)

http://iaeme.com/Home/journal/IJARET 100 [email protected]

 Dishang D. Trivedi, Urmil B. Bhatt and Santosh C. Vora

@
N  > 0O O *  O O +  O  O *   O   N (15)

Apart from the state and measurement equations, input vector K (Eqn. 12) is presented in

 N
discrete form as,


2J N 5  L  O M
 N
O
(16)

3.2. Simulation preliminaries and DSE of DFIG using EKF

3.2.1. Simulation aspects

For a considered SMIB system (Fig. 1), details of machine specification is provided in
Appendix B. Total simulation time is 6 s. Dynamic condition is simulated by considering
demand reduction in active power from wind generator. The demand is reduced by 40% at an
instance t = 2s. Sudden drop in required output active power warrants for change in reference
active power. Direct power control (DPC) scheme is employed to adapt to new power
requirements from grid. RSC control circuit adapts to change its output in accordance to new
power requirement from grid. In this work, synchronized operation of GSC and RSC control
circuit is considered to adjust to dynamic conditions. Output active power from DFIG to grid
is changed from 5.69 MW to 3.4 MW and output reactive power is changed from 0.3293 to
0.3986 MVAR as shown in Fig. 3. It is important to note that the reduction in power delivery
is met by the reduction observed in the stator and rotor output both, as limited control is offered
by the rotor circuit.

Figure 3. (a) Change in output active and (b) change in output reactive power under dynamic
condition

     and

With reduced power output condition, RSC control loop causes proportionate change in

viz.    &'( !

other current parameters viz. and voltage parameters

3.2.2. EKF implementation for DFIG

Normally active power and reactive power are measured with the help of current transformer
(CT) and potential transformer (PT) on AC side and with the help of sensors on converters.

http://iaeme.com/Home/journal/IJARET 101 [email protected]

 Application of Ekf Based Dynamic State Estimation For Dfig Rotor Power Control Under Faulty
Current Sensor Measurements

Due to unforeseen hostile conditions of wind generator, the possibilities exist where

corrupted with noise. To simulate such a noisy measurement environment, measurements 

measurements obtained from these instrument transformers attached to wind generator are

and Q are corrupted with Gaussian noise having covariance of 0.0001 p.u as shown in Fig.4.
Process noise covariance is 0.0001 p.u.

EKF algorithm is employed to estimate quantities for rotor control circuit viz.   and  
Figure 4 (a) Active power and power with noise (b) reactive power with and without noise

as well as stator currents  and  . During DSE process using EKF, and  forms
measurement matrixE. EKF works perfectly as DSE tool and it estimates all four dynamic state
viz.     and  of DFIG accurately as shown in Fig. 5 and Fig. 6 and hence, validating
accuracy of EKF algorithm.

Figure 5 (a)  actual and  estimated using EKF algorithm (b)  actual and   estimated
using EKF algorithm

http://iaeme.com/Home/journal/IJARET 102 [email protected]

 Dishang D. Trivedi, Urmil B. Bhatt and Santosh C. Vora

Figure 6 (a)  actual and  estimated using EKF algorithm (b)  actual and   estimated
using EKF algorithm

4. CASE STUDIES AND DISCUSSION

In DFIG,  and   , rotor currents are sensed using current sensors and then used for control
4.1. Case I: Current sensor measurement with noise

of active and reactive power output. Post load changes at t = 2 s, the output of current sensor is

The decoupled measurements  and   contains Gaussian noise having covariance of 0.01
assumed to be corrupted by noise at t = 2.3 s. This simulates faulty current output from sensor.

p.u. Effect of noise on measurements of  and  are depicted in Fig.7 (a) and Fig. 8(a)!

Figure 7 (a) Effect of noise on   (b) noisy  is replaced with 

http://iaeme.com/Home/journal/IJARET 103 [email protected]

 Application of Ekf Based Dynamic State Estimation For Dfig Rotor Power Control Under Faulty
Current Sensor Measurements

Figure 8 (a) Effect of noise on  (b) noisy  is replaced with  
Effect of noise in   and  deteriorate performance of control loop and hence, their effect
on rotor active power  and reactive power   is seen in Fig. 9(a) and Fig. 10(a). Noisy power
output from rotor also affects total power output of DFIG (Eqn. 10).
The moment at which effect of noise is observed in output of current loop and consequently

 and  are replaced with output of EKF based estimation algorithm  and  as shown

on active and reactive power at t = 2.3 s, in a short while (t = 2.32 s) output of current sensors

in Fig. 7(b) and Fig. 8(b)! From t = 2.32 s onwards estimated currents   &'(Uompared


with  and   (Fig. 2) to produce controlled rotor output  and  to desired level in RSC
control circuit. With estimated current inputs of   and   , nature of output rotor active and


noise free active power  and reactive power  (Eqn.10).

reactive power is observed as shown in Figs. 9(b)-10(b) respectively. This consequently causes

Figure 9 (a) Effect of noisy  and  on  (b)  after noisy currents replaced with  and
 

http://iaeme.com/Home/journal/IJARET 104 [email protected]

 Dishang D. Trivedi, Urmil B. Bhatt and Santosh C. Vora

Figure 10 (a) Effect of noisy  and  on  (b)  after noisy currents replaced with  and
 

4.2. Case II: Current sensor measurement with outliers

in measurements by current sensors (  &'(VW'UWX' and  ) as shown in Fig. 11(a) and

Simulating another possibility, the realistic condition is tested where random outliers show up

Fig. 12(a) respectively.

Figure 11 (a) Effect of outliers on  (b)   with outliers replaced with  

http://iaeme.com/Home/journal/IJARET 105 [email protected]

 Application of Ekf Based Dynamic State Estimation For Dfig Rotor Power Control Under Faulty
Current Sensor Measurements

Figure 12 (a) Effect of outliers on   (b)   with outliers replaced with  

The abrupt change in  due change in  and   is shown in Fig. 13(a).
Due to outliers in measurement signals, rotor active and reactive power change abruptly.

To detect the unwanted outliers, method of normalized innovation ratio [9] is proposed.
Readers are encouraged to refer Appendix C for overview of method of normalized innovation
ratio to detect outliers.

Figure 13 (a) Effect of  and   having outliers on  (b)  after measurement with outlers
replaced with  and  

existence of outliers at any instant  in output of current sensor. In presented case, once outlier
To overcome effect of outliers in measurement, algorithm (Appendix C) detects for

measurement from sensor (which are input to rotor control circuit) is replaced with   and 

is detected by the estimation algorithm at t = 3 s, immediately at t = 3.001 s current

instead of  thereafter, as shown in Fig. 11(b) and Fig.12(b) respectively.

Replacement with estimated variables   and  gives smooth and error free control of DFIG.


rotor active power  as indicated in Fig. 13(b). Results endorse that EKF based estimator
Once erroneous values are replaced with accurate output of EKF algorithm, it provides desired

values can be used as back-up for better control of DFIG when current sensors used in RSC
control circuit are infected with outliers. Similar abrupt effect of measurement outliers and

http://iaeme.com/Home/journal/IJARET 106 [email protected]

 Dishang D. Trivedi, Urmil B. Bhatt and Santosh C. Vora

 
removal of unstable output using   and  has been observed on other variables viz. 
!
,

4.3. Case III: Measurement unavailability due to current sensor failure

As another possibility, taking a worst case scenario following transient load change, a current
sensor fails and hence measurement to RSC control circuit is not available. This conditions is
simulated by taking random periods of measurement unavailability.

faulty current sensor. Faulty current sensor condition causes measurement unavailability in 
The measurement to RSC control circuit is considered to be erratically unavailable due

and  for period 0.5 s from t = 2.3 s to t = 2.5 s and subsequently for period of 1.4 s i.e. from
t = 3.8 s till t=5.2 s as shown in Fig. 14(a) and Fig. 15(a) respectively.

Figure 14 (a)  measurement unavailability for 0.5 s and 1.4 s duration (b) unavailable
measurements are replaced with  

Figure 15 (a)  measurement unavailability for 0.5 s and 1.4 s duration (b) unavailable
measurements are replaced with  

http://iaeme.com/Home/journal/IJARET 107 [email protected]

 Application of Ekf Based Dynamic State Estimation For Dfig Rotor Power Control Under Faulty
Current Sensor Measurements

Figure 16 (a) Effect of   and   measurement unavailability on  (b)  when unavailable

measurements are replaced with   and  

1) Observe rotor current measurement  output of current sensor at Y instance

To overcome this condition following algorithm is proposed:

3) If measurement is available, obtain decoupled rotor currents  and. Give input of
2) Check for measurement availability?

 and   to control loop and go to step no. 6.

4) If measurement  is not available due to current sensor fault, then use EKF estimated
value 
 and !


5) Compare estimated signals   and with  and  respectively.
  

6) Observe control loop for active and reactive power control.

Proposed algorithm first checks availability of measurement and if it recognizes the

provided with the output of EKF algorithm viz.   X'Z[W&(\]  X'Z[W&(\]  until

unavailability of measurement (in this case at time t=2.3 s), input to RSC control circuit is

input from current measurement resumes (in this case at time t = 2.8 s) as shown in Fig. 14(b)
and Fig. 15(b) respectively. Results indicate that estimated values obtained using EKF based
DSE faithfully fills the measurement void existed due to faulty current sensor operation. Hence,

measurement data unavailability. As seen in Fig. 16 (a), absence of measurements  
EKF based DSE works as a substitute to actual measurement from current sensor during

when current sensor measurement are replaced with    as shown in Fig. 16(b). It

also affects active power output of rotor. This discontinuity in rotor power is also overcome

assures that during unavailable measurement condition, continuity in active and reactive power

estimated values obtained using EKF. Similar effect of  and  unavailability and their
from rotor and consequently total active and reactive power can be maintained utilizing

replacement with     is also observed on other variables like  &'(. Results


prove ability of EKF based estimated variables can be used as back-up for current sensor issues.

5. CONCLUSION
DFIG based wind power has made it imminent for an EMS operator to have knowledge of its
dynamic states for better power control. After obtaining accurate dynamic state estimation of
DFIG, it can be used further to provide better control of DFIG, mainly for controlling its rotor
active and reactive power. Inherent nature of continuous dynamic operation of DFIG,
sometimes causes mal-operation and / or faulty operation of current sensor. The work presented
demonstrates successful application of estimated dynamic states achieved using EKF to

http://iaeme.com/Home/journal/IJARET 108 [email protected]

 Dishang D. Trivedi, Urmil B. Bhatt and Santosh C. Vora

current sensor which measures  =  &'( ? to RSC control circuit is simulated by
provide robust control under different faulty conditions of current sensor. Faulty conditions of

considering three conditions viz. noisy measurement output, measurements with outliers and
measurement unavailability. In all three conditions, the moment abnormality is detected in

quantities 
 and  resulted from EKF algorithm. Results indicate use of proposed novel

current sensor measurements, actual measurements are replaced by faithfully estimated

scheme which employs dynamically estimated states instead of faulty measurements and
successfully overcomes disadvantage of undesired output from DFIG due to ambiguous input
to control circuit under current sensor malfunctions.

REFERENCES
[1] B. Singh and S. Singh, “Wind power interconnection into the power system: A review of grid
code requirements,” The Electricity Journal, vol. 22, no. 5, pp. 54 – 63, 2009.
[2] Feijo J. Cidrs and C. Carrillo, “A third order model for the doubly fed induction machine,”
Electric Power Systems Research, vol. 56, no. 2, pp. 121-127, Nov. 2000.
[3] J. Ekanayake, L. Holdsworth and N. Jenkins, “Comparison of 5th order and 3rd order machine
models for doubly fed induction generator (DFIG) wind turbines,” Electric Power Systems
Research, vol. 67, no. 3, pp. 207 – 215, 2003.
[4] Y. Mishra, S. Mishra, F. Li, Z. Y. Dong and R. C. Bansal, “Small-signal stability analysis of a
DFIG-based wind power system under different modes of operation,” IEEE Transactions on
Energy Conversion, vol. 24, no. 4, pp. 972–982, Dec 2009.
[5] F. Wu, X. P. Zhang, K. Godfrey and P. Ju, “Small signal stability analysis and optimal control
of a wind turbine with doubly fed induction generator,” IET Generation, Transmission
Distribution, vol. 1, no. 5, pp. 751–760, September 2007.
[6] F. Wu, X. P. Zhang, P. Ju and M. J. H. Sterling, “Decentralized nonlinear control of wind
turbine with doubly fed induction generator,” IEEE Transactions on Power Systems, vol. 23,
no. 2, pp. 613–621, May 2008.
[7] A.Moualdia, MO.Mahoudi, L.Nezli and O.Bouchhida, “Modelling and Control of a Wind
Power Conversion System Based on the Double-Fed Asynchronous Generator”, International
Journal of Renewable Energy Research, Vol.2, No.2, 2012.
[8] L. Yang, Z. Xu, J. Stergaard, Z. Y. Dong, K. P. Wong and X. Ma, “Oscillatory stability and
eigenvalue sensitivity analysis of a DFIG wind turbine system,” IEEE Transactions on Energy
Conversion, vol. 26, no. 1, pp. 328–339, March 2011.
[9] L. Yang, Z. Xu, J. Ostergaard, Z. Y. Dong and K. P. Wong, “Advanced control strategy of
DFIG wind turbines for power system fault ride through,” IEEE Transactions on Power
Systems, vol. 27, no. 2, pp. 713– 722, May 2012.
[10] Dimitrios G. Giaourakis, Athanasios Safacas and Savvas Tsotoulidis, “Dynamic Behaviour of
a Wind Energy Conversion System Including Doubly-Fed Induction Generator in Fault
Conditions”, International Journal of Renewable Energy Research, Vol.2, No.2, 2012.
[11] Gadouche Zoubir, Belfedal Cheikh , Allaoui Tayeb and Belabbas belkacem, “Speed-Sensorless
DFIG Wind Turbine for Power Optimization Using Fuzzy Sliding Mode Observer”,
International Journal of Renewable Energy Research, Vol.7, No.2, 2017.
[12] Kai Rothenhagen and Friedrich Wilhelm Fuchs, “Doubly Fed Induction Generator Model-
Based Sensor Fault Detection and Control Loop Reconfiguration”, IEEE Trans. on Industrial
electronics, vol. 56, no. 10, Oct. 2009, pp-4229-4237.
[13] Shahram Karimi, Arnaud Gaillard, Philippe Poure and Shahrokh Saadate, “Current Sensor
Fault-Tolerant Control for WECS with DFIG”, IEEE Trans on Industrial electronics, vol. 56,
no. 11, Nov. 2009, pp-4660-4670.

http://iaeme.com/Home/journal/IJARET 109 [email protected]

 Application of Ekf Based Dynamic State Estimation For Dfig Rotor Power Control Under Faulty
Current Sensor Measurements

[14] Hui Lia, Chao Yanga, Yaogang Hua, Meng Zhaoa, Bin Zhaoa, Zhe Chenb, Shengquan Liua
and Dong Yang, ”Model and Performance of Current Sensor Observers for a Doubly Fed
Induction Generator,” Vol . 42, No. 10, pp-:1048–1058, 2014.
[15] N. K. A. Khedher and M. F. Mimouni, “Wind energy conversion system using DFIG controlled
by back stepping and sliding mode strategies,” Int. J. Renew. Energy Res., vol. 2, no. 3, pp.
421-430, 2012.
[16] M. T. EL-Hagry and M. N. Eskander, “Estimation of rotor voltage vector on the double excited
induction machine used in WECS, Electric Machines & Power Systems, 25, 8, pp. 839–850,
1997.
[17] R. Fan, Z. Huang, S. Wang, R. Diao and D. Meng, “Dynamic state estimation and parameter
calibration of a DFIG using the ensemble Kalman filter,” in 2015 IEEE Power Energy Society
General Meeting, July 2015,pp. 1–5.
[18] M. K. Malakar, P. Tripathy and S. Krishnaswamy, “State estimation of DFIG using an extended
Kalman filter with an augmented state model,” in 2014 Eighteenth National Power Systems
Conference (NPSC), pp. 1–6, Dec 2014.
[19] J. K. Mandal, A. K. Sinha and L. Roy, Incorporating nonlinearities of measurement function in
power system dynamic state estimation, IEE Proceedings - Generation, Transmission and
Distribution, vol. 142, no. 3, pp. 289–296, May 1995.
[20] Z. Huang, K. Schneider, J. Nieplocha and N. Zhou, Estimating power system dynamic states
using extended Kalman filter, in 2014 IEEE PES General Meeting — Conference Exposition,
pp. 1–5, July 2014.
[21] N.Sreekanth and Dr M.N.GiriPrasad, Comparative Ber Analysis of Mitigation of ICI Through
Sc,Ml and Ekf Methods In OFDM Systems, Volume 3, Issue 2, July- September (2012), pp.
69-86, International Journal of Electronics and Communication Engineering & Technology.
[22] Dishang D. Trivedi, Santosh C. Vora, Performance Assessment of EKF and UKF Based
Dynamic State Estimators for Intermittent Measurement Data in Power System International
Journal of Advanced Research in Engineering and Technology, 8(2), 2017, pp. 24–41
[23] D. Simon, Optimal State Estimation: Kalman, H-infinity and Nonlinear Approaches, Wiley
Publishers, 2006. (Book).

http://iaeme.com/Home/journal/IJARET 110 [email protected]