Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Machine Translated by Google

Renewable Energy 76 (2015) 662e669

Contents lists available at ScienceDirect

Renewable Energy
journal homepage: www.elsevier.com/locate/renene

Digital proportional multi-resonant current controller for improving grid-


connected photovoltaic systems
Pedro M. Almeida a, *, Pedro G. Barbosa a, Janaina G. Oliveira a, b,
Jorge L. Duarte
Paulo F. Ribeiro c
a
Power Electronics and Automation Group, Electrical Engineering Program, Federal University of Juiz de Fora, Juiz de Fora, MG, 36.036-900, Brazil b
Department of Electrical Engineering, Eindhoven University of Technology, Eindhoven, 5600, MB, The Netherlands Electrical
c
Engineering Institute, Federal University of Itajuba, Itajub a, MG, 37500-903, Brazil

article info abstract. abstract

Article history: This paper presents the modeling and design steps of a digital proportional multi-resonant controller used
Received March 1, 2014
in a grid-connected photovoltaic (PV) system. It is shown that the use of only one Proportional-Resonant
Accepted November 27, 2014 (PR) compensator, tuned to the system fundamental frequency, may have its effectiveness compromised
Available online 12 December 2014
due to nonlinearities in the system components. To overcome this drawback and improve the system's
output current waveform, a multi-resonant controller is introduced. The performance of the discrete-time
Keywords:
designed controller is tested on a grid-connected photovoltaic power plant. Experimental results obtained
Digital controller
with the operation of a 30 kWp PV system connected to a distribution network, using only the leakage
Resonant compensator
Current controller inductances of a connection transformer as passive filter are presented and discussed to demonstrate the
Grid-connected performance of the designed control strategy. © 2014 Elsevier
Photovoltaic Ltd. All rights reserved.
Discrete-time modeling

1. Introduction ground leakage current and DC current injection into the grid. On the other
hand, when galvanic isolation is mandatory, topologies based on
The integration of small and medium capacity sustainable sources- transformers must be used. Although high frequency transformers
based distributed generation systems into the electric grid is a fast and eliminate the ground leakage currents, they do not prevent DC current
environmentally friendly way to meet growing electricity demand in injection [3].
developed and developing countries. Based on the fact that the IEC 61727 [4] and IEEE 1547 [5] standards
Among renewable energy sources, photovoltaic (PV) systems are one of set limits for the maximum allowable amount of DC current which can be
the most promising, with a robust and exponential growth [1]. classically injected into the grid, the solution with line frequency transformer
between the converter and the grid will be used in this work. This solution
Several converter topologies have been proposed to implement the in addition to pre-venting the DC current injection, ensures galvanic
inverter and interface it with the grid [2]. They can be classified as single- isolation and voltage level adjustment.
phase or three-phase, with one or more stages. Another categorization
divides the topologies into three major classes: (i) Low-frequency Fig. 1 shows a PV system single line diagram. It consists of PV arrays
transformer coupling; (ii) High-frequency transformer; and (iii) connected to a voltage source converter (VSC) input terminals, along
Transformerless topologies. with acquisition system, signal conditioning, and con-trols. The interface
between the VSC and the grid is done only by a coupling transformer,
Transformerless topologies have the advantage of higher efficiency where its leakage inductances are used to attenuate the switching
and reduced size. However, they have issues with respect to frequency current harmonics. This approach eliminates the output low-
pass filter hence reducing the system size and cost.

Several strategies are used to control the interface converter.


* Corresponding author. Historically, the first control methods developed were in the
E-mail address: pedro.machado@engenharia.ufjf.br (PM Almeida).

http://dx.doi.org/10.1016/j.renene.2014.11.087
0960-1481/© 2014 Elsevier Ltd. All rights reserved.
Machine Translated by Google

PM Almeida et al. / Renewable Energy 76 (2015) 662e669 663

Fig. 1. Schematic diagram of a grid-tied PV system.

natural or abc reference frame [6]. They use nonlinear controllers (eg
R
hysteresis controllers) and have the disadvantages of requiring high
first

2
0 0 3 2
00 3
sampling frequencies and to switch the converter semi-conductors 6
L 7
L 7

with a variable frequency. d iaðatÞ 6

R
7
iaðtÞ
first
7
vaðtÞ
2 3
0 0 2 3
0 2
6 7 7

ibðtÞ
¼
6 7

ibðtÞ þ 66606666 7
vbðtÞ
As an alternative, the synchronous reference frame can be used. DT 6
L 7
L 7

4 icðtÞ 5
6 7

4 icðtÞ 5
7

4 vcðtÞ 3 5;
In the dq0 frame, the variables are referred to a coordinate system 6

R 7

first
7

synchronized with the grid AC signals [7]. The advantage of this 00 00


4 L5 4 L5
strategy is that the signals in the synchronous frame are no longer
(first)
AC, but DC signals. This allows the design of simpler linear controllers
(eg proportional-integral (PI)) [8]. However, a Phase Locked-Loop where ia(t), ib(t), and ic(t) are the instantaneous currents at the
(PLL) is required to estimate the angle of the grid voltages in order to converter's terminals; vaðtÞ¼ðvt;a ðtÞ va;pcc ðtÞ=NÞ, vbðtÞ¼ðvt;b
ensure the correct transformation [9,10]. ðtÞ vb;pcc ðtÞ=NÞ, and vcðtÞ¼ðvt;c ðtÞ vc;pcc ðtÞ=NÞ; vt;a ðtÞ;
vt;c ðtÞ, and vt;c ðtÞ are the instantaneous voltages at the con-
Another approach is the use of the stationary ab0 reference frame.
verter's AC terminals; va;pcc ðtÞ; vb;pcc ðtÞ, and vc;pcc ðtÞ are the
This strategy, when applied to a three phase, three-wire system instanta-neous voltages at the point of common coupling (PCC); N is
reduces the number of controllable variables [6,11]. How-ever, as the transformer turns ratio; L and R are the equivalent leakage induc-
with the natural reference frame, the currents and voltages are time tance and the series resistance of the connection transformer windings
varying waveforms. reflected to the primary side, respectively.
This paper describes the design of a digital proportional multi- Applying the zero-order hold method, as explained in Appendix
resonant current controller. The objective is to compensate dis-torted A [12], to discretize (1), yields.
currents generated by a photovoltaic system, especially when the PV The discretization of (1) by using the zero-order hold (ZOH)
system is connected to the grid without output pas-sive filters. method (Appendix A) yields,

Initially an analytical model is presented in the abc frame and then iaðan þ iaðnÞ vaðtÞ
it is discretized and transformed to an ab0 frame. The aforementioned 2
1Þ ibðn þ
3 ¼F 2
ibðnÞ
3
þG 2
vbðtÞ (2)
model is used to design the inner current com-pensators. When the 4 1Þ icðn þ 1Þ
5 4 icðnÞ 5 4 vcðtÞ 3 5;
system is connected to the grid, with only a Proportional-resonant
(PR) controller tuned to the fundamental frequency, distorted currents where (n þ 1) and (n) represents the discrete time tnþ1 ¼ (n þ 1)Ts
are generated due to the nonlinear behavior of the connection and tn ¼ nTs, respectively; n is a positive integer that represents the
transformer. With the design and the inclusion of additional resonant sampling time and Ts is the sampling time. The matrices F and G are
controllers, the distorted current is compensated. Experimental results calculated by
are presented to demonstrate and validate the performance of the
digital pro-portional multi-resonant controller.
2 RTs 3
6

6
e L
0 0 7

6 7


6 7

0 e
RTs
L
0
7
; (3)
2. AC-side model 6

6
7

6 7

6 7

RTs

Based on Fig. 1, and ignoring the high frequency switching 0 0 e L

4 5
harmonics, the following system can be written for the converter's AC
side
Machine Translated by Google

664 PM Almeida et al. / Renewable Energy 76 (2015) 662e669

and,
iaðn þ 1Þ ¼ 4iaðnÞ þ duaðnÞ ibðn
(9)
þ 1Þ ¼ 4ibðnÞ þ dubðnÞ :
2 RTs 3
L
6

1e 7

In this new system, the dynamic behavior of ia(n þ 1) and ib(n þ 1)


6

0 0 7

6
R 7

can be directly controlled by ua(n) and ub(n), respectively.


6 7

6 7

The a-axis current control block diagram is depicted in Fig. 2, where


6
RTs 7

G¼ 6

first e L 7

:
(4) the modulation signal ma(n) is used to generate pulse-width modulation
6

0 0 7

(PWM) signals and, therefore, to control the output current. The control
R
6 7

6 7

6
7

7
strategy is based on a negative feedback loop where the actual current
6
RTs
L
7

ia(n) is compared with a reference i aðnÞgenerating an error ÿa(n)


e
6 7

first
6

0 0
7

which is processed by a controller C(z) generating a control signal


4
R 5 ua(n) .
Note in Fig. 2 that the control signal ua(n) and the voltage 3 p .
Using the Clarke transform [8] to convert (2) from abc frame to ab 0 ffiffiffi

ðva;pcc ðnÞ=NÞ are added before they are divided by VDCðnÞ= This
frame, the following state-space representation can be written
procedure, besides normalizing the modulation signal ma(n) due to the
fact that a space vector modulation (SVM) is used [13], has a
iðan þ 1Þ iaðnÞ vaðnÞ
2 3 ¼ G 2 ibðnÞ 3
þH 2
(5) feedforward action reducing the influence of the variations of grid
ibðn þ 1Þ vbðnÞ
voltage and minimize the effect of the DC voltage's ripple on the current
4 i0ðn þ 1Þ 5 4 i0ðnÞ 5 4 v0ðnÞ 3 5;
controller's output signal. The delay z1 is introduced in the block
where the subscripts a, b and 0 identify the currents and voltages in the diagram to model the processing time. A similar control block is used to
new coordinate frame; vaðnÞ¼ðvt;a ðnÞ va;pcc ðnÞ=NÞ, vbðnÞ¼ðvt;b control the b-axis current.
External loops may be included to control the level of the DC voltage
ðnÞ vb;pcc ðnÞ=NÞ and v0ðnÞ¼ðvt;0 ðnÞ v0;pcc ðnÞ=NÞ. The
matrices G and H are given by and the reactive power injected into the grid [14]. The design of the
aforementioned controllers will not be addressed in this paper.
400
2
G ¼ ½T
first
¼
040 (6) Based on the fact that the signals to be controlled on ab frame are
4 004 3 5; sinusoidal, proportional-resonant controllers can be used to ensure a
zero steady state tracking at its resonance frequency [15].
and,

d00
H¼ 2323 first
¼ 2 0d0 (7) 3.1. Current controller design
4T 55½T
4G 4
00d 3 5;
The transfer function of a continuous-time proportional-reso-nant
, d ¼are
1 where 4 ¼ eðRTs=LÞ (1 the Clarke's
4)/R; direct and inverse transformation
T and [T] (PR) controller is given by Ref. [16]
matrices (Appendix B) [7,8].
uðsÞ S
CPRðsÞ ¼ ¼ kp þ CrðsÞ ¼ kp þ kr eðsÞ ; (ten)
3. Control strategy s2 þ u2 r

Using (5) and referring to the three-phase three-wire circuit shown where kp and kr are the proportional and resonant gains, respec-tively;
in Fig. 1, the following relations can be written for the voltages at the Cr(s) the transfer function of the resonant controller; and ur is the
converter's AC terminals resonant frequency. Fig. 3 shows the block diagram of a PR controller
given in (10) based on two integrators.
8 In order to digitally implement the PR controller, (10) must be
vt;a ðn þ 1Þ ¼ uaðnÞ þ va;pcc ðnÞ . N vt;b
; (8) disclosed. There are several ways to discretize controllers with Euler's
<
: ðn þ 1Þ ¼ ubðnÞ þ vb;pcc ðnÞ . N and Tustin integration methods being very popular when proportional-
integral (PI) controllers are used. In the case of PR controllers, these
where ua(n) and ub(n) are the new control variables in the ab reference low-order methods are not suitable because they cause a displacement
frame. The inclusion of the terms ðva;pcc ðnÞ=NÞ and ðvb;pcc ðnÞ=NÞ of the resonant poles position [17]. This characteristic, in turn, can
in (8) has a feedforward action reducing the influence of the grid voltage drastically reduce the controller's effectiveness for higher resonance
on the current control loop [8]. frequencies.
Substituting (8) into (5) and neglecting the row and columns related According to [17] the first-order hold (FOH) is one of the con-version
to zero sequence current and voltage yields. methods which provides a precise pole location and good

Fig. 2. a-axis current controller block diagram.


Machine Translated by Google

PM Almeida et al. / Renewable Energy 76 (2015) 662e669 665

Fig. 3. Block diagram of PR controller based on two integrators.

delay compensation. Therefore, applying the FOH method, as


shown in the Appendix C, to Cr(s) given in (10) yields.

1 cosðurTsÞ 1z2 _
CrðzÞ ¼ kr (11)
u2r Dr " 1 2cosðurTsÞz1 þ z2 # :

Consequently, the discrete-time version of the PR controller is.


Fig. 4. Nyquist diagram of G(z) and kp$G(z), for kp ¼ 2.66 V/A.
CPRðzÞ ¼ kp þ CrðzÞ: (twelfth)

There are several methodologies to design the PR controller. Most According to [20] reasonable values for the vector margin are
of them are based on gain and phase stability margins aided by between 0.50.8. By doing so, it ensures a safe distance from the
Bode's diagram. Although these methods work quite well for critical point and as a result a well damped transient
systems in which the frequency response decay monotonically, response, allied to good switching and high frequency
they are not suitable for more complex systems where the gain and attenuation.
phase cross curves 0 dB and 180, respectively, more than once [18]. Based on the above, and with h ¼ 0.7 the corre-sponding proportional
For systems in which the open loop has reactive poles, the gain is kp ¼ 2.66 V/A. Fig. 4 depicts the
vector margin should be used. The vector margin is defined to be Nyquist diagram of G(z) and kp$G(z) for the system parameters
the distance to the (1,j0) point from the closest approach of the given in Table 1.
Nyquist plot. In other words, it is the inverse of the maximum After the value of the proportional gain is obtained, the con-troller's
sensitivity function value. Because the vector margin is a single resonant portion can be included and the influence of kr on
margin parameter, it removes all the ambiguities in assessing stability that the system's stability analyzed in order to choose an appropriate
comes with using gain and phase margins in combina-tion [18]. gain.gain.

Fig. 5 depicts the Nyquist plot of G(z)CPR(z) for different values of


The sensitive function is given by kr. As can be seen, increasing the value of the resonant gain results
in a reduction of the stability margin. For kr ¼ 3000 V/A$s the
first
vector margin is reduced from 0.7 to 0.665, which is still a
SðzÞ ¼ ; (13)
1 þ GðzÞCPRðzÞ reasonable distance from the critical point.
The choice of kr ¼ 1000 V/(A$s) provides a good tradeoff be-tween
where G(z) is the plant's transfer function, which based on (9) is transient response and selective filtering [19]. This value also
given by yields a good slope of the Nyquist curve at the crossover frequency,
which drastically affects the performance and robustness of the
dz2 closed-loop system [21].
GðzÞ ¼ z1 iaðzÞ ¼ z1 ibðzÞ ¼
(14)
uaðzÞ ubðzÞ 1 4z1 ;
4. Experimental results
first

where the delay z is incorporated into the plant's transfer function.


As previously defined, the distance of the open-loop transfer The current controller designed in the previous section and
function G(z)CPR(z) to the critical point (1,j0) at each frequency u, which parameters are shown in Table 2, were implemented using a
for z ¼ ejuTs , in the Nyquist diagram is fixed-point TMS320F2812 DSP (digital signal processor) [22].
Print Fig. 6, the instantaneous currents injected into the AC grid
DðzÞ ¼ 1 þ GðzÞCPRðzÞ; (15) in steady state are shown. It can be seen that these currents
contains some harmonic components which is verified by the
where DðzÞ is the inverse of the sensitivity function. Therefore, the
vector margin is defined as h ¼ minjDðzÞj, which is a dimensionless
quantity. quantity. Table 1
Based on the fact that the responsible part of the PR controller only System parameters.
changes the frequency response in a small range around the reso-nance Value
description
frequency, the proportional gain can be designed first. After
Switching frequency (fsw) 20 kHz
that, the influence of the responsive controller may be included and
Sampling frequency (fs) 12 kHz
its effects analyzed in order to establish a suitable value for kr. Fundamental frequency (f1) 60 Hz
For a given vector margin, the correspondent kp gain can be Phase rms voltage (Vt) 127 V
found easily through Nyquist plot graphical analysis aided by a Leakage inductance (L) 0.83 mH
Series resistance (R) 0.37 U
computer. An analytical relation between these two parameters can
Turns ratio (N) 2
also obtained [19]. However, it involves complex calculations.
Machine Translated by Google

666 PM Almeida et al. / Renewable Energy 76 (2015) 662e669

Fig. 7. Harmonic spectrum of the “a” phase current, THD ¼ 8.5%.

5. Multi-resonant controllers
Fig. 5. Nyquist diagram of G(z)CPR(z) for different values of kr, and kp ¼ 2.66 V/A.

The harmonic components present in the appearance appear can be


Table 2 compensated using several resonant controllers tuned to the fre-quency
PR controller parameters. of each individual harmonic to be attenuated.
description Value Based on the fact that the resonant controller presents a high
gain at the resonant frequency and a small influence on the
Proportional gain (kp) 2.66 (V/A)
remaining frequency spectrum, this characteristic allows the ag-
Resonant gain (kr) 1000 V/(A$s)
Vector margin (h) 0.692 gregation of several resonant controllers together, as shown in
Fig. 8.
The resulting transfer function of the association of several
harmonic spectrum depicted in Fig. 7. The total harmonic responsive controllers in parallel is
distortion (THD) is 8.5%. Note that the magnitude of the fifth and
seventh harmonics, as well as the THD, are above the limits
recommended by Ref. [5]. Cr;tðzÞ ¼ kp þXnh Cr;hðzÞ; (16)
H
As the converter is connected directly to the interface trans-former's
terminals, the voltage with high-frequency components
where h is the harmonic order and nh is highest frequency har-monic
forces the transformer to operate in a non-linear region. Conse-quently,
component.
it distorts the output currents.
In order to avoid this problem, the line transformer should be As kp affects the frequency spectrum equally, only one propor-
tional gain is used. Therefore, it can be calculated as shown in the
oversized or passive filters can be connected between the static
Section 3.1.
converter and transformer. In both cases there is an increase in
According to Fig. 7 at least two responsive controllers must be
volume and cost of the PV system. The next section provides security
included, tuned to the 5th and 7th harmonics, in order to guarantee
alternative solution for this problem based on multiples resonant
controllers. that the appearance currently complies with the standard [23]. Even
Although the 11th and 13th harmonics are below the specified limits,
resonant controllers tuned to these frequencies are added as well to
improve the quality of the generated power and the current THD.

Fig. 6. Three-phase currents at the converter's terminals (horiz. scale: 2 ms/div; vert. Fig. 8. Block diagram of a Proportional Multi-resonant controller tuned to funda-mental,
scale: 22 A/div). 5th, 7th, 11th and 13th harmonics.
Machine Translated by Google

PM Almeida et al. / Renewable Energy 76 (2015) 662e669 667

Fig. 9 shows the Nyquist diagram of G(z)Cr,t(z) with the inclusion


of the resonant controllers tuned to 5th, 7th, 11th and 13th har-
monics. It should be noted that the system presents two gain
margins, one positive and one negative (smaller). This does not
imply instability, because there is no clockwise encircles of the
(1,þj0) point. Although it means that the system is conditionally
stable. It may still become unstable if the gain is increased or
decreased. Fact which is not desirable.
Despite the fact that the system is stable, the vector margin is
too small (h ¼ 0.098), which means that the control loop has a poor
stability margin and also that the response time may present poor
damped oscillations. Another important issue is that the processing
delay may affect the controller performance or even cause insta-
ability [15].
In order to overcome the drawbacks, the following delay
compensation was proposed in Ref. [24]

s cosðkurTsÞ ur sinðkurTsÞ Fig. 10. Nyquist diagram of G(z)Cr,t(z) with delay compensation, for kr ¼ 1000 V/(A$s),
Cd
r ðsÞ ¼ ; (17)
kr s2 þ u2 r and kp ¼ 2.66 V/A.

where k is the number of sampling periods to be compensated.


According to [25], k ¼ 2 results in a suitable compensation value. in parallel with the previously designed PR controller. The additional
Applying the FOH method to discretize (17) gives controllers are tuned to 300, 420, 660 and 780 Hz. The

2z1 sinðkurTsÞ½sinðurTsÞ cosðurTsÞ


CD
r þ (18)
u2r Ts þ
ðzÞ ¼ kr ( 1 z1 ½cosðkurTsÞ 1 sinðkurTsÞurTs
2z1cosðurTsÞcosððk
þ z2 þ 1ÞurTsÞ u2r Ts 1 2z1 cosðurTsÞ þ z2 ) :

Fig. 10 depicts the Nyquist diagram of G(z)Cr,t(z) with the in- resonant gain kr of each new compensator is 1000 for the reasons
clusion of the delay compensation for the 11th and 13th harmonics. explained above.
The 5th and 7th do not need compensation. Fig. 11 shows the currents at the converter's terminals with the
As can be clearly seen, the delay compensation drives the inclusion of the additional resonant compensators. It can be seen
Nyquist trajectory far from the instability point. As a result, the vector that the currents are sinusoidally shaped. The harmonic compo-
margin increased from 0.098 to 0.485, which is a reasonable stability nents at which the resonant controllers are tuned to, have their
margin. amplitude reduced as shown in Fig. twelfth.
A comparison between Figs. 7 and 12 show that the output
5.1. Experimental results currents are significantly improved achieving hence a suitable

In order to improve the waveforms of the current injected into the


grid, four resonant controllers, for each axis (a and b), are added

Fig. 11. Three-phase currents at the converter's terminals with multiple resonant controllers
Fig. 9. Nyquist diagram of G(z)Cr,t(z), for h ¼ 1,5,7,11 and 13, and kr ¼ 1000 V/(A$s), and (horiz. scale: 2 ms/div; vert. scale: 22 A/div).
kp ¼ 2.66 V/A.
Machine Translated by Google

668 PM Almeida et al. / Renewable Energy 76 (2015) 662e669

analysis and frequency response through the Nyquist diagram and


vector margin. Experimental results show the effectiveness of the
approach.

Acknowledgments

The authors would like to thank CAPES, CNPq, FAPEMIG,


INERGE, UFJF, TU/e and UNIFEI for financial support, scholarship
and for providing the resources to test the PV system and the
proposed control strategy.

Appendix A. Zero order hold equivalent of the state-space


equation

Fig. 12. Harmonic spectrum of the “a” phase current, THD ¼ 2.14%. The discrete-time solution of the state-space equation x_ðtÞ ¼
AxðtÞ þ BuðtÞ is given by Ref. [twelfth]

eAðtnþ1tÞ BuðtÞdt; (A.1)


xðtnþ1Þ ¼ eAðtnþ1tnÞ xðtnÞ þ Ztnþ1
tn

where tn and tnþ1 are two successive sampling instants of a


continuous time system.
Assuming u(t) is held constant in a sampling interval Ts ¼ (tnþ1
tn), that is, uðtÞ ¼ uðtnÞ; c tn t<tnþ1, it is possible to rewrite (A.1) as,

2 3
xðtnþ1Þ¼½eATs xðtnÞ þ 6 eAðtnþ1tÞ Bdt uðtnÞ;
7
(A.2)
Ztnþ1
4 tn 5

where,

F ¼ eATs ; (A.3)

and,
Fig. 13. Voltage and current at the point of common coupling (horiz. scale: 2 ms/div;
vert. scales: CH1111 A=div and CH2125 V=div).

eAðtnþ1tÞ Bdt: (A.4)


G ¼ Ztnþ1
power quality indicator according to [23]. The currents' THD is tn

drastically reduced from 8.5% to 2.14%.


Fig. 13 shows the voltage and the current at the point of common Note that F and G as A and Ts are constants. Thus, starting the
coupling. sampling operations at t0 ¼ 0, and making use of an uniform
sampling period Ts, (A.2) can be rewritten as shown bellow,
6. Conclusion
xðn þ 1Þ ¼ F xðnÞ þ G uðnÞ; (A.5)
This paper presents the discrete-time modeling and control design
of a three-phase, single stage grid-connected photovoltaic system. where (n þ 1) ¼ (n þ 1)Ts and n ¼ nTs denote the discrete time tnþ1
The system is controlled by a current-mode controller where digital and tn, respectively.
proportional multi-resonant controller is used. The controller design
is based on the discrete-time analytical model of the interface Appendix B. The Clarke's transformation
converter. It is shown that a PR compensator tuned to the fundamental
frequency is not sufficient to ensure that the converter synthesizes The direct and inverse transformation of the voltages and cur-
sinusoidal currents as a result of the non-linear operation of the rents between the abc to ab0 reference frames, and vice-versa, are
connection transformer. This undesired behavior could have been given by Refs. [7,8],:
avoided by using passive filters be-tween the converter's terminals
and the coupling transformer. (B.1)
fab0ðnÞ ¼ T fabcðnÞ;
However, in addition to increasing volume and cost of the PV system,
these filters may cause resonance issues if many of those systems
are connected to the power distribution network. Alter-natively, and,
several resonant controllers are introduced to compen-sate the
distortion present in the output currents. This solution does not lead
fabcðnÞ ¼ T1fab0ðnÞ; (B.2)
to any change in the hardware originally developed.
The design of the responsive controller is based on the stability
Machine Translated by Google

PM Almeida et al. / Renewable Energy 76 (2015) 662e669 669

where, References

first first
[1] EPIA. Global market outlook for photovoltaics 2013e2017. Tech. rep. Euro-
2
first

3
pean Photovoltaic Industry Association; May 2013.
2 2
7 [2] Teodorescu R, Liserre M, Rodríguez P. Grid converters for photovoltaic and
ffiffiffi ffiffiffi

2
7

wind power systems. 1st ed. Wiley-IEEE Press; 2011.


3p 3p 7

T¼ 66660666
7
; (B.3) [3] Franceschini G, IEEE EL, Buticchi G. Saturation compensation strategy for grid
3 2 2 connected converters based on line frequency transformers. IEEE Trans Energy
7

first 1 1
7

Conversations 2012;27(2):229e37.
[4] IEC 61727. Photovoltaic (pv) systems and characteristics of the utility interface.
42 2 2 5 Tech. rep. IEC; 2004.
[5] IEEEe1547. IEEE std. 1547e2003 standard for interconnecting distributed
resources with electric power systems. Tech. rep. IEEE; 2003.
and, [6] Buso S, Malesani L, Mattavelli P. Comparison of current control techniques for
active filter applications. IEEE Trans Power Electron 1998;45(5):722e9.
[7] Krause PC, Wasynczuk O, Sudhoff SD. Analysis of electric machinery and drive
1 01
systems. 2nd ed. Wileye IEEE Press; 2002.
[8] Yazdani A, Iravani R. Voltage-sourced converters in power systems e
2 3
ffiffiffi

6
first 3p 2 7
modeling, control, and applications. IEEE Press; 2010.
6 first 7
[9] Chung SK. A phase tracking system for three phase utility interface inverters.
T1 ¼ 2 ; (B.4) IEEE Trans Power Electron 2000;15(3):431e8.
6 7

6 7

ffiffiffi

6 7

[10] Rodriguez P, Teodorescu R, Candela I, Timbus A, Liserre M, Blaabjerg F. New


first 3p
first positive sequence voltage detector for grid synchronization of power con-
4 2 2 5 verters under faulty grid conditions. In: Proc. power electronics specialist conf;
2006. p. 1e7.
[11] Akagi H, Watanabe EH, Aredes M. Instantaneous power theory and its application
where fabcðnÞ¼½faðnÞ fbðnÞ fcðnÞ0 , fab0ðnÞ¼½faðnÞ fbðnÞ f0ðnÞ0 , and, to power conditioning. 1st ed. Wiley-IEEE Press; 2007.
the subscript ð0Þ denotes matrix transpose. The factor (2/3) is introduced in (B.3) to [12] Moudgalya K. Digital control. 1st ed. WileyeInterscience; 2007.
[13] Buso S, Mattavelli P. Digital control in power electronics. 1st ed. Morgan and
guarantee amplitude invariance be-tween ðabcÞ/ðab0Þ quantities.
Claypool Publishers; 2006.
[14] Yazdani A, Fazio ARD, Ghoddami H, Russo M, Kazerani M, Jatskevich J, et al.
Modeling guidelines and a benchmark for power system simulation studies of
three-phase single-stage photovoltaic systems. IEEE Trans Power Del
2011;26(2):1247e64.
Appendix C. The FOH discretization of a continuous compensator [15] Yuan X, Merk W, Stemmler H, Allmeling J. Stationary-frame generalized in-
tegrators for current control of active power filters with zero steady-state error
for current harmonics of concern under unbalanced and distorted operating
conditions. IEEE Trans Ind Appl 2002;38(2):523e32.
The first-order hold (FOH) method assures that the response of discrete equivalent [16] Zmood DN, Holmes DG, Bode G. Frequency domain analysis of three-phase
linear current regulator. IEEE Trans Ind Appl 2001;37(2):601e10.
system for a ramp input will be roughly equal to the output of continuous sampling
[17] Yepes AG, Freijedo FD, Doval-Gandoy J, Lopez Oscar, Malvar J, Fernandez-
system at the instants. Comesaa P. Effects of discretization methods on the performance of resonant
In this way, assuming C(z) is the equivalent discrete form of the continuous controllers. IEEE Trans Power Electron 2010;25(7):1692e712.
[18] Franklin GF, Powell JD, Emami-Naeini A. Feedback control of dynamic sys-
compensator given by the transfer function C(s), to guarantee the ramp invariance
tems. 6th ed. Prentice Hall; 2009.
for the time domain output, the following equality should be satisfied [26] , [19] Yepes AG, Freijedo FD, Lopez Oscar, Doval-Gandoy J. Analysis and design of
resonant current controllers for voltage-source converters by means of nyquist
diagrams and sensitivity function. IEEE Trans Ind Electron 2011;58(11):5231e50.

Tsz first
[20] Astrom KJ, Murray RM. Feedback systems: an introduction for scientists and
Z
first

CðsÞ$ (C.1) engineers. 2007.


s2 t¼nTs ;
1 (CðzÞ$ ðz 1Þ 2)¼L [21] Karimi A, Garcia D, Longchamp R. PID controller tuning using bode's integrals.
IEEE Trans Control Syst Technol 2003;11(6):812e21.
[22] http://www.ti.com/lit/ds/symlink/tms320f2812.pdf.
where L are the
1 and Z
Laplace and z inverse transformation, respectively, and Ts is the
first

[23] IEEE. IEEE std.928e1986 recommended criteria for terrestrial photovoltaic power
systems. Tech. rep. 1986.
sampling period.
[24] Bojoi RI, Griva G, Bostan V, Guerriero M, Farina F, Profumo F. Current control
Then, apply the z-transform in both sides of (C.1) yields. strategy for power conditioners using sinusoidal signal integrators in syn-
chronous reference frame. IEEE Trans Power Electron 2005;20(6):1402e12.
[25] Limongi LR, Bojoi R, Griva G, Tenconi A. Digital current-control schemes. IEEE
1 z12 z1Ts first
first
Ind Electron Mag 2009;3(1):20e31.
CðzÞ ¼ $ZL CðsÞ$ (C.2)
s2 t¼nTs : [26] Chen CT. Analog and digital control system design: transfer-function, state-
space, and algebraic methods. Saunders College Pub; 1993.

You might also like