Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

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.2018.2870407, IEEE
Transactions on Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS

Real-Time Wavelet-Based Grid Impedance


Estimation Method
Denis K. Alves, Student Member, IEEE, Ricardo L. A. Ribeiro, Member, IEEE,
Flavio B. Costa, Member, IEEE, and Thiago O. A. Rocha

Abstract—This paper presents a novel method to es- Active methods that employ a signal injection to estimate
timate the grid impedance based on stationary discrete the grid impedance overcome the passive methods drawbacks
wavelet packet transform (SDWPT). The proposed method [6]. Their implementation can follow transient or steady-state
uses a steady-state technique, by injecting an interhar-
monic current into the grid and measuring the voltage techniques [2]. The transient methods employ noncharacteris-
response at the point of common coupling (PCC) to esti- tic signal injections as voltage disturbance, or current spike to
mate the grid impedance. The proposed method employed estimate the grid impedance over a wide frequency range via
a standard three-phase photovoltaic (PV) system intercon- a postprocessing procedure [7]. This complex signal analysis
nected to the grid to validate its effectiveness experimen- generates much more data than needed and can overload the
tally. Comparisons with a discrete Fourier transform (DFT)-
and continuous wavelet transform (CWT)-based impedance real-time control algorithm of power interfaces. An alternative
estimation approaches demonstrate the performance of is to use the steady-state approach based on a periodic inter-
proposed method. Besides, the proposed SDWPT-based harmonic signal injection and analyzing the system response
impedance estimation provided accurate experimental re- as proposed in [2].
sults, which make it viable for real-time applications.
Both transient and steady-state active methods usually use
Index Terms—Grid impedance estimation, PV grid-tied the discrete Fourier transform (DFT). However, its perfor-
inverters, stationary discrete wavelet packet transform.
mance can be affected by interharmonic distortions and tran-
sient events. The use of continuous wavelet transform (CWT),
with a complex mother wavelet, instead of DFT demonstrated
I. I NTRODUCTION
to be an alternative to overcome these limitations [8]. However,
the CWT is not usually embedded for real-time applications
R ENEWABLE energy sources interconnect the grid via
power converters acting as active front-end (AFE) in-
terfaces for regulating the output power flow [1]. This new
due to its complexity and high redundancy of the provided
information.
configuration demands efforts to improve the AFEs control The discrete wavelet transform (DWT) versions, such as
strategies to avoid undesirable grid instabilities. Therefore, the the SDWPT, with real mother wavelet have been used as
knowledge of grid impedance is essential to proper control and alternative to the CWT successfully in several practical ap-
operation decisions and its implementation as an embedded plications. However, DWT versions with real mother wavelet
strategy is useful for the actual power system model [2]. present mathematical limitations for estimating phase angle
Different methods estimate grid impedance through passive [9]. Based on this statement, the DWT has been claimed
or active approaches [2]. Passive approaches use information to be unsuitable for impedance estimation [8]. However, the
of noncharacteristic voltage and current harmonics already use of cross analysis of both voltage and current system
present in the system to estimate the grid impedance. Their response overcomes the lack of phase information, and makes
implementation can use: controlled excitation of LCL based the SDWPT-based impedance estimation totally possible, as
inverter [3], recursive least-square technique [4], extended demonstrated in this paper.
Kalman filter [5], or discretized grid-tied model in a rotating
reference frame analysis [6]. These estimation schemes depend This paper introduces a real-time SDWPT (RT-SDWPT)
on the background voltage and current distortions, and present with a real mother wavelet for estimating the grid impedance
low signal-to-noise ratio (SNR), which results in weaker considering the steady-state active strategy. As advantages,
estimation [6]. the SDWPT presents both the uniform frequency bandwidth
and has the time-invariant property. RT-SDWPT employs the
injection of an irrelevant interharmonic far from low-order
Manuscript received April 10, 2018; revised July 04, 2018 and August harmonics to avoid power quality problems. Accurate experi-
10, 2018; accepted September 03, 2018. This work was supported by mental results obtained in a PV generation platform intercon-
the Brazilian National Council for Scientific and Technological Develop-
ment (CNPq). nected to the grid validated the proposed real-time grid esti-
D. K. Alves, R. L. A. Ribeiro, F. B. Costa, and T. O. A. Rocha mation method, and the embedded RT-SDWPT demonstrated
are with the Department of Electrical Engineering, UFRN, Natal, its practical applicability. Besides, the proposed SDWPT-based
RN, 59072-970, Brazil; e-mail: alvesk3@hotmail.com, rlucio@ct.ufrn.br,
flaviocosta@ect.ufrn.br and toar37@hotmail.com. impedance estimation provided accurate experimental results,
which make it viable for real-time applications.

0278-0046 (c) 2018 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.2018.2870407, IEEE
Transactions on Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS

II. T HE WAVELET-B ASED G RID I MPEDANCE E STIMATION The reactance of the grid impedance is determined by a
The PV voltage source inverter (VSI) injects an interhar- simple mathematical procedure as follows:
monic in the reference current of its standard current control
Xgw (k) = Xfwint (k)ω1 /ωfint , (7)
loop. The PV control system uses the most generic current
control mode approach, based on a multi-loop scheme, in where ωint and ω1 are frequencies of the interharmonic
which the inner loop refers to the current regulation. Therefore, and fundamental. Therefore, the grid impedance estimation is
the interharmonic injection corresponds to modify the standard given by
reference current by adding the interharmonic synthesized q
(if int ) by the estimation algorithm. The method considers Zgw (k) = (Rgw (k))2 + (Xgw (k))2 . (8)
that there is no actual current interharmonic component in the
frequency selected to add the synthetic current interharmonic.
III. E XPERIMENTAL R ESULTS
The frequency of the interharmonic (fint ) must be higher
than the system fundamental frequency (f ) and low-order Fig. 1 shows the block diagram of the PV laboratory setup
harmonics for avoiding disturbances on the PV operation. The employed for realizing experimental tests. It comprises of an
SDWPT sampling frequency must attend two criteria: 1) the 8 kWp PV three-phase grid-tied inverter interconnected to an
harmonic components (odd and even harmonics) are at the actual 15 kVA power grid. A dSPACE executes the standard
cutoff frequencies of the wavelet packet decomposition, and control algorithm of the PV system.
2) the choice of interharmonic frequency corresponds to the The RT-SDWPT grid impedance estimation technique is
center of one of wavelet sub-bands. also embedded in a dSPACE and employs a sampling fre-
fint
The spectral energy of the interharmonic (Ej,x ) is esti- quency of fs = 1920 Hz (32 cycles of the fundamental f =
mated, in the real-time, with the wavelet coefficient energy 60 Hz). Four-level decomposition (j = 4) was necessary to
of the signal x, at scale j, which include the interharmonic include all harmonics in the cutoff frequencies. The selected
frequency fint in the band center, as follows: interharmonic frequency is fint = 630 Hz, centered in the
k
X Xh
L−1 i2 wavelet band [600 - 660] Hz far from the fundamental and
fint
Ej,x (k) = sfj int (n − l) , (1) low-order harmonics. The grid impedance estimation uses the
fint fint
n=k−N +1 l=0 wavelet coefficients energy variables E4,V and E4,I based
where k ≥ N -1, s is the wavelet packet coefficients of x on the mother wavelet of the Daubechies family with 30
at scale j, and N is the number of samplings per cycle of coefficients (vanishing moments of 15).
the power system frequency (f = 50 or 60 Hz). The discrete
PV array LCL
time-domain root mean square (RMS) interharmonic voltages
dc-link Filter ig PCC
and currents can be described in terms of the wavelet packet a lg rg vg
coefficients as follows [10]: VSI igb
q Boost lg rg vg
fint
Xf int (k) = Ej,x /2N , (2) converter igc lg rg vg

where X = V for voltage or X = I for current; vgabc


At the actual sampling k, the phase angle between power
system voltage and current of the interharmonic (θf int ) and the
interharmonic impedance (Z bfint ) can be properly computed by Control PWM
using the RT-SDWPT-based approach as follows: vga ZSDWPT
ifint Hz PV Control
 k
"L−1 #
1 X X fint fint
Display
Ej,v (n − l)Ej,i (n − l)  Injected vg System
 2N

 abc ZDFT
n=k−N+1 l=0
θfint (k) = cos−1 
 
Vfint (k)Ifint (k)

 
  Fig. 1. Block diagram of the 8 kWp PV three-phase laboratory setup.

(3)
The proposed RT-SDWPT grid impedance estimation was
bfint (k) = (Vfint (k)/Ifint (k))6 θfint . similar to the DFT- and CWT-based methods such as shown
Z (4) in Fig. 2 and 3. Red dotted lines present the average values
obtained via DFT method and black dashed ones represent the
bfint (k) refers to the resistance of the grid
The real part of Z average values obtained via offline analysis by the CWT in two
impedance: intervals: before and after grid impedance changes, whereas
time-domain graphs show the grid impedance estimation via
bfint (k)} = Vfint (k) cos(θfint (k)).
Rgw (k) = Re{Z (5) the SDWPT approach in the real-time.
Ifint (k)
Fig. 2 depicts the experimental results (1 second) for
The reactance related to the interharmonic frequency corre- grid resistance step changing by adding external resistors of
sponds to the imaginary part of Zbf int (k) given by
Re = 0.5 Ω in series with the grid. Initially, the proposed
method estimated the standard grid impedance composed by
bfint (k)} = Vfint (k) sin(θfint (k)).
Xfwint (k) = Im{Z (6)
Ifint (k) Rgw ≈ 0.6 Ω and Xgw ≈ 0.15 Ω (according to the grid

0278-0046 (c) 2018 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.2018.2870407, IEEE
Transactions on Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS

1.5 Raverage
Raverage 1.5
Grid resistance (Ω)

1.2

Grid resistance (Ω)


1.2
Grid resistance Increase
0.9 Grid resistance increase 0.9 0.5 Ω
0.5 Ω
0.6
0.6
SDWPT Grid resistance after SDWPT
0.3 DFT Raverage Raverage Grid resistance after added
added a resistor 0.3 DFT
CWT CWT a resistor and inductance
0.0 0.0
(a)
0.18 (a) Xaverage
Grid reactance (Ω )

0.40
0.15

Grid reactance (Ω)


0.12 Xaverage 0.32
Grid reactance Increase
0.09 Grid reactance 0.24 0.21Ω
Grid reactance after
0.06 0.16
added a resistor
0.03 Xaverage Grid reactance after added
0.08
0.0 a resistor and inductance
(b) 0.0
1.5 (b)
Z average Z average
1.5
Grid impedance (Ω)

1.2

Grid impedance (Ω)


1.2
0.9 Grid impedance increase Grid impedance Increase
0.5 Ω 0.9 0.70Ω
0.6
Z average Grid impedance after 0.6
0.3 Z average Grid impedance after added
added a resistor
0.3 a resistor and inductance
0.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Time (s) 0.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
(c) Time (s)
(c)

Fig. 2. Experimental results of grid impedance estimation with the re-


sistance step changing: (a) grid resistance estimation; (b) grid reactance Fig. 3. Experimental results of grid impedance estimation with the induc-
estimation and; (c) grid impedance estimation. tance step changing: (a) grid resistance estimation; (b) grid reactance
estimation and; (c) grid impedance estimation.

transformer nameplate). When the controlled switches inter-


connected the external resistors (Re ), the proposed method obtained from a PV laboratory setup, compared with DFT and
estimated a resistance around Rgw ≈ 1.1 Ω that corresponds CWT estimation methods, demonstrated the feasibility of the
to the predicted resistance of composed grid impedance. proposed RT-SDWPT estimation method.
Fig. 3 depicts the experimental test for grid inductance
step changing by inserting external inductors of Le = 0.5
mH, with internal resistances of Rgw ≈ 0.63 Ω, in series R EFERENCES
with the grid. When the controlled switches interconnected
[1] H. Liu, P. C. Loh, X. Wang, Y. Yang, and W. W. D. Xu, “Droop control
the external inductors, the proposed method estimated values with improved disturbance adaption for a pv system with two power
around Rgw ≈ 1.3 Ω and Xgw ≈ 0.37 Ω, which corresponds conversion stages,” IEEE Trans. Ind. Electron., vol. 63, no. 10, pp. 6073–
to the predicted values of grid impedance association. Both 6085, Oct. 2016.
[2] L. Asiminoaei, R. Teodorescu, F. Blaabjerg, and U. Borup, “Implementa-
experimental results with the RT-SDWPT-based method have tion and test of an online embedded grid impedance estimation technique
not used any kind of post-processing filter or an average value for pv inverters,” IEEE Trans. Ind. Electron., vol. 52, no. 4, pp. 1136–
to the estimation of the grid impedance. 1144, Mar. 2005.
[3] M. Liserre, F. Blaabjerg, and R. Teodorescu, “Grid impedance estimation
via excitation of lcl -filter resonance,” IEEE Trans. Ind. Appl., vol. 43,
IV. C ONCLUSION no. 5, pp. 1401–1407, Sept. 2007.
On the contrary of stated in the literature that discrete [4] S. Cobreces, E. J. Bueno, D. Pizarro, F. J. Rodriguez, and F. Huerta,
“Grid impedance monitoring system for distributed power generation
wavelet transforms with the real mother wavelet are unable electronic interfaces,” IEEE Trans. Instrum. Meas., vol. 58, no. 9, pp.
to estimate the grid impedance, this paper introduced an RT- 3112–3121, Sept. 2009.
SDWPT-based method for grid impedance estimation based on [5] N. Hoffmann and F. W. Fuchs, “Minimal invasive equivalent grid
impedance estimation in inductive resistive power networks using ex-
steady-state active approach. The proposed method presents tended kalman filter,” IEEE Trans. Power Electron., vol. 29, no. 2, pp.
easy implementation and is proper for the insertion in AFEs 631 – 641, Feb. 2014.
as an embedded algorithm. The proposed approach estimated [6] A. Ghanem, M. Rashed, M. Sumner, M. A. Elsayes, and I. I. I. Mansy,
“Grid impedance estimation for islanding detection and adaptive control
the grid impedance under step changing of resistance and of converters,” IET Power Electronics, vol. 10, no. 11, pp. 1279–1288,
inductance with reasonable accuracy. Experimental results Sept. 2017.

0278-0046 (c) 2018 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.2018.2870407, IEEE
Transactions on Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS

[7] M. Sumner, B. Palethorpe, and D. W. P. Thomas, “Impedance measure- new framework for complex wavelet transforms,” IEEE Trans. Signal
ment for improved power quality-part 1: the measurement technique,” Process., vol. 51, no. 7, pp. 1825 – 1837, Jul. 2003.
IEEE Trans. Power Del., vol. 19, no. 3, pp. 1442 – 1448, Jul. 2004. [10] D. K. Alves, F. B. Costa, R. L. A. Ribeiro, C. M. S. Neto, and
[8] M. Sumner, A. Abusorrah, D. Thomas, and P. Zanchetta, “Real time T. O. A. Rocha, “Real-time power measurement using the maximal
parameter estimation for power quality control and intelligent protection overlap discrete wavelet-packet transform,” IEEE Trans. Ind. Electron.,
of grid-connected power electronic converters,” IEEE Trans. Smart Grid, vol. 64, no. 4, pp. 3177–3187, Apr. 2017.
vol. 5, no. 4, pp. 1602 – 1607, Jul. 2014.
[9] F. C. A. Fernandes, R. L. C. V. Spaendonck, and C. S. Burrus, “A

0278-0046 (c) 2018 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.

You might also like