Sunteți pe pagina 1din 6

A step-by-step parameter design method for discrete

state space control scheme of dynamic voltage


restorer
He Lei, Rengcun Fang, Dongjun Yang, Jingyou Xu, Bofeng Gao, Yong Kang
Qiushi Xu, Zixia Sang State Key Laboratory of Advanced Electromagnetic
State Grid Laboratory for Hydro-thermal Power Resources Engineering and Technology
Optimal Allocation & Simulation Technology Huazhong University of Science and Technology
State Grid HBEPC Economic & Technology Research Wuhan, China
Institute
Wuhan, China
Lei Wu
SG Huangshi Electric Power Supply Company
Huangshi, China
Abstract virtual resistor and the Posicast controller [12-13]. From the
pole point of view, appropriate pole placement will achieve a
Abstract—Dynamic Voltage Restorer (DVR), as a device of high-performance control system. A single-loop control
voltage sag compensation, demands fast transient response, and method using constrained pole placement was proposed in
the disturbance of source voltage is a main impact on control [14]. However, due to shortage of freedom degrees, the
performance. The DVR control strategy based on discrete state
space will enhance control bandwidth and stability. However, for
imperfect pole-placement could yield lower control bandwidth,
the discrete space control system, the number of parameters is relatively. To avoid it, [2,15] proposed a control strategy
relatively more which will lead to a complicated design process. based on discrete state space, established the closed-loop
To solve this problem, this paper proposed a control parameter control system model which is combined with
design method based on the idea of virtual resistor. Firstly, by proportional-integral (PI) and discrete full state feedback, and
means of state feedback control, an adjustable virtual resistor the one-beat-delay (OBD) is feedback as a extra state variable.
can be series in filter inductor. According to the requirements of Such controller not only takes charge of reference tracking
steady-state error, phase margin and Routh’s criterion in w only, but also solved the decoupling component of load
domain, the variation regions of PI parameters and virtual voltage and grid voltage/load current.
resistance can be solved by a step-by-step method, and then the
appropriate control parameters can be selected. Experimental
Although the multi-feedback control including OBD
results from a digital signal processor-based system are included to feedback enables a fully controllable system, the numbers of
support theoretical analysis. unknown coefficients are relatively increased. A common
solution is to employ the direct pole placement, whereas the
Keywords—Dynamic voltage restorer; Discrete state space control; location of desired poles and zeroes is highly dependent on
Parameter design; Virtual resistor; Step-by-step experience [2,14-16]. To solve it, based on the state space
control mentioned earlier, this paper presents a parameter
I. INTRODUCTION design strategy by using the idea of virtual resistor [13]. The
With the high-precision load becomes more and more proposed state feedback control including load voltage, filter
widespread, the requirement of power quality has been current and inverter output voltage feedback is employed to
improved, even if short-time voltage sag will cause impact on emulate a physical resistor. Such control has an advantage of
the load. Dynamic voltage restorer (DVR), as a power avoiding from being affected by the system delay in discrete
electronic device to compensate voltage sag, has become a domain. Another merit is that adjusting a virtual resistance
heat area of investigation[1-2]. might be equivalent to adjusting three feedback coefficients.
For a DVR control system, voltage feedback controller In another word, the system order will be reduced. The
should be introduced to take charge of reference tracking, and remaining unknown coefficients including virtual resistance
the disturbance of source voltage should be suppressed and coefficients of PI regulator can be selected from a
effectively. Various control methods for a DVR have been satisfactory region without try and error. This region is
proposed in previous studies, including state feedback control obtained by a step-by-step method to meet the requirements of
[3-4], proportional resonant control [5], repetitive control the control system performance, including stable state error,
[6-7], predictive control [8], band controller [9], H∞ control phase margin and Routh’s criterion in w domain. Based on
[10-11], etc. Several active damping techniques are employed that, the proposed method can enable a high performance
to dampen the LC resonance in a converter system, such as a controller, which ensures a well-damped, highly robust, fast

l-))) 
transient-response and low steady-state-error system. The system delay (Td) can be modeled as the sum of the one
Experimental results are provided in section IV to verify the sampling period delay Ts and PWM transport delay 0.5Ts
validity of the proposed design method. [2,15]. The converter output voltage in z-domain can be
governed by the following equation,
II. DISCRETE STATE-SPACE CONTROLLER
Vi k  1 Vi* k (5)
Fig.1(a) shows the equivalent single-phase diagram of DVR,
Hence, the continuous and discrete generalized control plants
and the transformer leakage inductance is used as filter
are presented in Fig. 2, respectively.
inductance. Fig.1(b) shows the equivalent circuit of Fig.1(a),
Vi, Vs, and Vl represent the inverter output voltage, source Il Vs

voltage, and load voltage, respectively. If and Il represent the


filter inductor current and load current, respectively. Lt, rt and Vi* Vi x y
e sTd
Cf represent the leakage inductance and winding resistance of
the injection transformer, filter capacitance, respectively. The
turn’s ratio of the injection transformer is assumed to be 1:1. (a)
Il k Vs k

Vi* k Vi k x k y k

(b)
(a) Fig. 2. State-space model of a single-phase DVR in (a) continuous domain
and (b) discrete domain.

By considering Vi as an extra state variable and Vi* as a new


input variable, the system can be modeled as
(b) xc k  1 Gcxc k  H1cVi* k  H 2c I l k  H 3cVs k (6)
Fig. 1. (a) Simplified one-line diagram of DVR. (b) Equivalent circuit of where
DVR.
xc k ª¬ x k Vi k º¼
T ª G H1 º
Gc «0
¬ 1u2 0 »¼
Based on Fig. 1(b), the following equations can be derived
by applying Kirchhoff’s voltage and current laws as H1c ª¬0 1u2 1º¼
T
H 2c >H2 0@
T
H 3c > H3 0@
T

dI f Fig.3 shows the control block diagram of the discrete state


Vi Vl  Vs  rt It  Lt
dt (1) space controller, all the state are feedback through the vector
dVl
If Cf
dt
 Il
K cf ª KV K I KVi º «¬
(7)
»¼
l f
The continuous state space equation can be written as
x& Ax  Bu G zfi is the discretized load current feed-forward decoupling
(2)
y Cx path and G zfv is the discretized source voltage feed-forward
where decoupling path, which have been discussed detailedly in [2].
x ªVl
¬ I f º¼
T
u >Vi I l Vs @
T
The PI regulator is presented as
ª 0 1 º ª0 z Ki
 1 0º GPI Kp  (8)
« Cf » « Cf » C >1 0@ 1  z 1
A «
« 1
»
r »
B > B1 B2 B3 @ «
«1
»

«  t» «L 0 Il k Vs k
¬ Lt Lt ¼ ¬ t Lt »¼
G zfi G zfv
Equation (2) can be discretized as
x k  1 Gx k  Hu k
(3) Discrete
y k Cx k Vl* k z
Vi* k Generalized Vl k
GPI
Control Plant
where G and H are given by Vl k [Fig.2(b)]
x k
G e ATs H > H1 H2 H3 @
Ts
³0 G W dW B
Vi k Kcf
Taking into account the time delay introduced by digital
control, the inverter voltage can be expressed as
Vi Vi* * e sTd (4) Fig. 3. Control block diagram of the proposed controller.


This control system based on PI regulator and OBD with inductor in continuous domain is given as
feedback yields five control parameters. How to solve the ª 0 1 º
control parameters conveniently and accurately is a key issue. « Cf » (9)
Aˆ « »
The detailed derivation based on virtual resistor will be « 1  r »
« L L f »¼
discussed in the next section. ¬ f
where r is the desired virtual damping resistance. The discrete
system and input matrices can be calculated as
III. PARAMETER DESIGN BASED ON VIRTUAL ­Gˆ e AT ˆ
s
RESISTOR ° (10)
® T
°̄ Hˆ 1 ³0 s Gˆ W dW B1
According to above analysis, for PI+state space control
system, the number of parameters is relatively more which Hence, the discrete system and input matrices taking into
will lead to a complicated design process. Some literatures account the OBD can be calculated with
discuss the use of discrete direct-design pole placement to ­ ª Gˆ Hˆ 1 º
ˆc
achieve a high performance controller, but the desired °°G « » (11)
® ¬« 0 1u2 0 ¼»
closed-loop pole and zero locations are specified by °ˆ T

¯° H1c ª¬0 1u 2 1º¼


experience [2,14-16]. To solve this problem, an intuitive and
where
simple controller design method based on virtual resistor is
­Gˆ e AT
ˆ
proposed in this paper. A virtual resistor in series with the °
s
(12)
®
filter inductor or parallel with the filter capacitor can be Ts ˆ
°̄ H1 ³0 G W dW B1
ˆ
realized by using full state-feedback to further damp the LC According to the calculation above, the damped system model
resonance. Moreover, the state-feedback vector K’f can be in discrete time domain is given as
equivalently replaced by a virtual resistance, which means C Hˆ z 1 (13)
that the number of control parameters will be reduced. In Gˆ p z
zI 2  Gˆ
1

order to meet the specifications of the control system The poles of the system model can be obtained by using
performance, including steady-state error, phase margin and MATLAB, which is defined as
Routh’s criterion in w domain, a step-by-step method can be (14)
used to derive the entire parameter regions. Thus, the Pˆ eig ªGˆ z º ¬ p ¼
appropriate control parameters can be obtained without try The feedback vector can be calculated with direct pole
and error. placement, which is written as
A. Implementation of the idea of virtual resistor Kˆ c ac ker Gˆ c, Hˆ c, Pˆ (15)
f 1
Considering the inductor current is already measured in the
state-space control system, a virtual resistor in series with the Therefore, the relation between feedback vector and
filter inductor is preferred [13]. The active-damping method damping resistance can be established. The state-space control
based on single inductor current feedback is no longer system is approximately equivalent with a dual-loop control
equivalent to a resistance but a frequency-dependent system, leading to a simple parameter design. It’s worth
impedance due to the OBD in digital control. The mentioning that, although the poles of the system model by
characteristics of equivalent impedance will vary with using full state-feedback can be the same with the model by
frequency, which might lead to a hard control design. To employing a damping resistor, but the zeros are different in
avoid this, full state-feedback including converter output discrete time domain.
voltage, inductor current and load voltage is implemented, The discrete system model by using full state-feedback is
which is shown in Fig.4. expressed as
ui* z ui z ul z Gp z
1  cos ZTs z  1 (16)
1 1 2  aT
ˆ 2 aT
ˆ
z 1
sL f Cf s
3
z  2z e s
cos Zˆ Ts  ze s

where aˆ r , Zˆ Z  aˆ , the natural frequency Zr


2 2 1 .
Kui r
it z 2L f Lf C f
Ki f
ul z The discrete system model by employing a damping resistor is
Kul defined as
§ 1  aˆ e  aT
ˆ s
sin ZˆTs  e aTs cos Zˆ Ts ·¸ z  §¨ e 2aTs  a e aTs sin Zˆ Ts  e aTs cos Zˆ Ts ·¸
ˆ ˆ ˆ ˆ ˆ
¨
Gˆ p z © Zˆ ¹ © Zˆ ¹
3 2  aTˆ s 2 aT
ˆ s
ui* z z  2z e cos ZˆTs  ze
ui z ul z
z 1
1
sL f
1
Cf s
(17)
The equation (16) is divided by (17), and the transfer function
r is written as
Gp z 1  cos ZTs z  1
Fig. 4. Implementation of the idea of virtual resistor. Gze z
Gˆ z §1  aˆ e aT
ˆ s
sin Zˆ Ts  e aTs cos Zˆ Ts ·¸ z  §¨ e2 aTs  a e aTs sin Zˆ Ts  e aTs cos Zˆ Ts ·¸
ˆ ˆ ˆ ˆ ˆ
p ¨
© Zˆ ¹ © Zˆ ¹
Assuming the system matrix of a damping resistor in series (18)


The bode plot of equation (18) is shown in Fig.5, where it can ZcTs 2

2
Ki d ª1  Z Z 2 º  rZ C (24)
be seen that the magnitude and phase of Gze(z) will vary a lot K ze «¬ c r »
¼ c f

only in high frequency range. As Gze(z) has a relatively slight (2) Steady-state error
phase-lag beyond resonant frequency, it is approximately The control system should have higher open-loop gain at
equivalent with the proportional gain Kze for simplicity, which fundamental frequency in order to minimize the steady-state
is specified as follows: error. With (21), the gain To can be represented as
1  cos ZTs (19) ­ K T
2 ½
(25)
K ° °
ze
1  e 2 aT
ˆ s
 2e  aT
ˆ s
cos Zˆ Ts To
°
20 log10 ®
K ze K p2  i 2 s
Zo °
¾ t Tlim
° ª1  Zo Zr 2 º  rZoC f 2
2
Thus, the approximately equivalent continuous system model °
° ¬ ¼ °
by using full state-feedback can be obtained as ¯ ¿

K ze e sTd
where, Tlim is the desired gain at fundamental frequency. The
G s
(20)
p 2
L f C f s  rC f s  1
integral gain of PI regulator Ki can be solved from (23) and
(25), yielding
T

 rZ C º»¼  ª«¬ 1 Z  rZ C º»¼


Bode Diagram lim
ª 2 2 2 2 2
Ts 10 10 * « 1  Zo Zr Zr
5 2
Zr
Ki t
¬
o f c c f
(26)
1 Zo2 1 Zc2
Magnitude (dB)

K ze
0

(3) Phase margin


Larger phase margin will bring less overshoot. The phase
-5
5 margin can be derived from (21) as
§ K T ·
K zee sTd ¨ K p  i s ¸
Phase (deg)

0 © s ¹
PM S ‘
L f C f s 2  rC f s  1 s jZc

-5
2 3 4 5 § rC f · §K T · (27)
S  1.5ZcTs  arctan ¨ ¸  arctan ¨ i s
10 10 10 10
¸ t PM lim
Frequency (rad/sec)
¨ 1  Z Z 2 ¸ ¨ Zc K p ¸
© c r ¹ © ¹
Fig.5. Bode plot of the transfer function Gze. where, PMlim is the desired phase margin. The integral gain of
PI regulator Ki can be solved from (23) and (27), yielding
B. Control parameter design
The coefficients of the PI regulator and the damping
resistance should to be specified in order to meet the Ki t
ZcTs 1  Z c Zr
2 2

 rZcC f
2

requirements of steady-state error, phase margin and Routh’s


K ze 1 
1  Zc Zr  1.5Zc2Ts rC f  tan PM lim 1.5ZcTs 1  Zc Zr
2
2
 rZ C
c f

stability criterion. According to the analysis above, the


1.5ZcTs 1  Zc Zr
2

 rZcC f  tan PM lim 1  Zc Zr  1.5Zc2Ts rC f
2

open-loop transfer function from the voltage reference to the (28)
load voltage with the proposed control strategy in continuous (4) Routh’s stability criterion
domain can be expressed as It’s worth mentioning that Routh’s stability criterion is not
§ K T · suitable for discrete system, thus, the closed-loop control
K ze e sTd ¨ K p  i s ¸
§ Ki Ts · © s ¹ system needs to be mapped from z domain to w domain by
Gop s ¨ Kp  ¸ * Gp s (21)
© s ¹ L f C f s 2  rC f s  1 using w transform. The discrete control system closed-loop
(1) Cutoff frequency transfer function is
A high cutoff frequency ωc will offer a relatively high G pi z G p z
Gcl z (29)
control bandwidth, which is typically slightly beyond natural 1  G pi z G p z
frequency ωr. To ensure that the open-loop control system has The denominator polynomial in z domain is proposed from (8),
unit gain at cutoff frequency, the magnitude calculation of (21) (16) and (29) as
gives D z m0 z 4  m1 z 3  m2 z 2  m3 z  m4 0 (30)
K ze K 2p 
Ki Ts
2
where,
Zc2 (22)
Gop jZc 1 ­m0 1
2
°
2


ª1  Z Z 2 º  rZ C  aT
ˆ
«¬ c r »
¼ c f
°m1  2e s cos ZˆTs  1
With (22), the proportional gain of PI regulator Kp can be °° (31)
obtained as ®m2 e2aTs  2e aTs cos Zˆ Ts  1  cos ZrTs K p  Ki
°
ˆ ˆ

2 °m3 1  cos ZrTs Ki  e2aT
ˆ s

2
ª1  Z Z 2 º  rZ C
«¬ c r »
¼ c f Ki Ts
2
(23) °
Kp 2
K ze

Zc2
°̄m4 1  cos ZrTs K p
The mapping z=(1+w)/(1-w) is used to provide the
In order to make (23) valid, the integral gain of PI regulator Ki
denominator polynomial in w domain, such that
should be written as
D w a0 w4  a1w3  a2 w2  a3 w  a4 0 (32)


where, A 20 kVA three-phase experimental prototype is built to
­ a0 1  e 2aTˆ s
 2e  aTs cos Zˆ Ts
ˆ verify the proposed control parameter design method. The
° experimental parameters of the DVR are listed in Table I.
° a1 3  e
°°
2 aT
ˆ s
 2e  aT
ˆ s

cos Zˆ Ts  1  cos ZrTs 2 K p  Ki (33) TABLE I. Experimental parameters of DVR
® a2 3  e
2 aT
ˆ s
 2e  aT
ˆ s

cos Zˆ T  1  cos Zr Ts 4 K p  Ki Grid voltage
°
° a3 1  e
°
2 aT
ˆ s
 2e  aT
ˆ s

cos ZˆT  1  cos ZrTs 2 K p  Ki Rated voltage
Rated frequency
380 V
50 Hz
°̄ a4 1  cos ZrTs Ki Transformer
Applying Routh’s stability criterion, the stable conditions for Turns ratio 220 V
the discrete control system are derived as follows Leakage inductance 1 mH
­a0 ! 0 VSI
°a ! 0
° 1 Rated power 20 kVA
° a1a2  a0 a3
!0
(34) Sampling frequency 20 kHz
°c13 Filter capacitor (Cf) 50 μF
® a1
° c13 a3  a1a4
°c14 !0
° c13 According to the parameters of the DVR system, the natural
°a ! 0
¯ 4 frequency ωr is equal to 4472 rad/s. Thus, the desired cutoff
As c13 and c14 are transcendental equations, it is complicated frequency ωc, the lower limit of gain at fundamental
to yield the relation between Ki and r from (34) numerically. frequency Tlim and phase margin PMlim are designed to be
An alternative solution is to obtain the stable region by using 4000 rad/s, 25 dB, and 45°, respectively. The virtual
ezplot(\cdot) function embedded in MATLAB. As a result, resistance r and the integral gain of PI regulator Ki are
Fig.6 shows the satisfactory region of Ki and r might be taken specified to be 13.5 and 0.2076 respectively from Fig.6,
to meet the requirements of stability from (24), (26), (28) and which is employed with the parameters shown in Table I and
(34). the desired values mentioned earlier. With (15), the
3 parameters of controller are obtained as
ª¬ K cf Kp K i º¼ > 0.7501 9.2486 0.4773 1.8413 0.2076@
2.5
(35)
Bode Diagram
2 Gm = 9.79 dB (at 1.04e+004 rad/sec) , Pm = 45.3 deg (at 4.39e+003 rad/sec)
50

1.5
Ki

Magnitude (dB)

1
-50

0.5 -100
0

0 -180
Phase (deg)

0 20 40 60 80 100
r -360

Fig.6. The reasonable region of Ki and r constrained by the requirements. -540

-720
2 3 4 5
10 10 10 10
It can be seen from Fig.6 that the appropriate values of Ki Frequency (rad/sec)

and r can be selected in the shaded area. Comparatively Fig.7. Bode plot of the open-loop transfer function from the voltage reference
speaking, the requirements of phase margin and steady-state to the load voltage with the proposed control strategy.
error have a relatively major impact on the satisfactory region.
For the virtual damping resistance r, the span of satisfactory Fig.7 shows the bode plot of the control to the output
range is too large. As a large r might lead to an overdamped open-loop transfer function for proposed controller. The cutoff
system, the gain of PI regulator has to be increased to improve frequency ωc, gain at fundamental frequency To and phase
the transient response. As a result, the control system will be margin PM are obtained to be 4390 rad/s, 25.3 dB, and 45.3°,
sensitive to high frequency disturbance, resulting in poor respectively. It’s worth mentioning that the obtained ωc is
robustness. In this case, it is necessary to select a relatively larger than the desired one. The reason is that compared with
smaller value of virtual resistance r. Furthermore, it should be the PI regulator discretized by backward differentiation
noted that larger Tlim and PMlim will introduce smaller shaded method, the gain of continuous PI regulator is slightly smaller,
area, increasing the lower limit of r. Hence, the value of Tlim mainly above the corner frequency. However, to simplify the
and PMlim might be selected as appropriate. calculation, the design of control parameters is based on
continuous PI regulator. As a result, a larger ωc is obtained in
IV. EXPERIMENTAL RESULTS the discrete control system.


The experiment is carried out by symmetrical three phase resistor. Based on that, the satisfactory regions of the
voltage sag, and the results are shown in Fig. 8. The DVR remaining unknown coefficients for meeting the specifications
restores voltage sags with 10%, 60%, and 100% from 40ms to of the system performance, including steady-state error, phase
100ms, respectively. Taking into account that the circuit and margin and Routh’s criterion in w domain, can be obtained by
control scheme for each phase are independent, hence, only the the proposed step-by-step method without try and error. The
a-phase voltage waveform is shown. The experimental results performance of the proposed control parameter design method
show that the proposed control parameter design method is has been verified by experimental results.
effective in the discrete state space control scheme.

REFERENCES
[1] A. M. Rauf and V. Khadkikar, "An enhanced voltage sag compensation
scheme for dynamic voltage restorer", IEEE Trans. Industrial
Electronics, Vol. 62, No. 5, pp. 2683–2692, May. 2015.
[2] H. Lei, M. Y. Xue, X. C. Lin, and Y. Kang, “A discrete state-space
control scheme for dynamic voltage restorer”, Journal of Power
Electronics, Vol. 13, No. 3, pp. 400-408, May. 2013.
[3] H. Kim and S. K. Sul, “Compensation voltage control in dynamic
voltage restorers by use of feed-forward and state feedback scheme”,
IEEE Trans. Power Electronics, Vol. 20, No. 5, pp. 1169-1177, Sept.
2005.
[4] P. T. Cheng, J. M. Chen, and C. L. Ni, “Design of a state-feedback
controller for series voltage-sag compensators”, IEEE Trans. Industry
(a) Application, Vol. 45, No. 1, pp. 260–267, Jan/Feb. 2009.
[5] M. J. Newman, D. G. Holmes, J. G. Nielsen, and F. Blaabjerg, “A
dynamic voltage restorer (DVR) with selective harmonic
compensation at medium voltage level”, IEEE Trans. Industry
Application, Vol. 41, No. 6, pp. 1744–1753, Nov/Dec. 2005.
[6] Y. O. Cho and S. K. Sul, “Controller design for dynamic vo1ltage
restorer with harmonics compensation function”, in Proc. IAS, pp.
1452-1457, 2004.
[7] P. R. Sánchez and E. Acha, “Dynamic voltage restorer based on flying
capacitor multilevel converters operated by repetitive control”, IEEE
Trans. Power Delivery, Vol. 24, No. 2, pp. 951–960, Apr. 2009.
[8] J. D. Barros and J. F. Silva, “Multilevel optimal predictive dynamic
voltage restorer”, IEEE Trans. Industrial Electronics, Vol. 57, No. 8,
pp. 2747–2760, Aug. 2010.
[9] S. Sasitharan and M. K. Mishra, “Constant switching frequency band
controller for dynamic voltage restorer”, IET Power Electronics, Vol.
(b) 3, No. 5, pp. 657–667, 2010.
[10] Y. W. Li, D. M. Vilathgamuwa, F. Blaabjerg, and P. C. Loh, “A robust
control scheme for medium-voltage-level DVR implementation”,
IEEE Trans. Industrial Electronics, Vol. 54, No. 4, pp. 2249–2261,
Aug. 2007.
[11] Y. W. Li, F. Blaabjerg, D. M. Vilathgamuwa, and P. C. Loh, “Design
and comparison of high performance stationary-frame controllers for
DVR implementation”, IEEE Trans. Power Electronics, Vol. 22, No. 2,
pp. 602–612, Mar. 2007.
[12] Y. W. Li, P. C. Loh, F. Blaabjerg, and D. M. Vilathgamuwa,
“Investigation and improvement of transient response of DVR at
medium voltage level”, IEEE Trans. Industry Application, Vol. 43, No.
5, pp. 1309–1319, Sept/Oct. 2007.
[13] Y. W. Li, “Control and resonance damping of voltage-source and
current-source converters with LC filters”, IEEE Trans. Industrial
Electronics, Vol. 56, No. 5, pp. 1511–1521, May. 2009.
[14] R. Turner, S. Walton, and R. Duke, “Robust high-performance inverter
(c) control using discrete direct design pole placement”, IEEE Trans.
Fig.8. Single phase voltage sag with different depths (Time: 10ms/div, Industrial Electronics, Vol. 58, No. 1, pp. 348–357, Jan. 2011.
CH1-CH2:200V/div). [15] M. Y. Xue, Y. Zhang, Y. Kang, Y. X. Yi, S. M. Li, and F. R. Liu, “Full
(a) Voltage sag depth 10%. (b) Voltage sag depth 60%. (c) Voltage sag depth grid voltage feed-forward for discrete state feedback controlled
100%. grid-connected inverter with LCL filter”, IEEE Trans. Power
Electronics, Vol. 27, No. 10, pp. 4234–4247, Oct. 2012.
V. CONCLUSION [16] J. Dannehl, F. W. Fuchs, and P. B. Rhgersen, “PI State Space Current
Control of Grid-Connected PWM Converters With LCL Filters”, IEEE
As PI+state feedback control leads to a high order system, a Trans. Power Electronics, Vol. 25, No. 9, pp. 2320–2330, Sept. 2010.
proper parameter design method has been employed to reduce
the system order from five to three by emulating a physical



S-ar putea să vă placă și