Distributed generation systems (DGSs) have been getting more and more attention in terms of renewable energy use and new generation technologies in the past decades. The selfexcited induction generator (SEIG) occupies an important role in the area of energy conversion due to its low cost, robustness and simple control. Unlike synchronous generators, the SEIG has to absorb capacitive reactive power from the outer device aiming to stabilize the terminal voltage at load changes. This paper presents a novel static VAR compensator (SVC) called a magnetic energy recovery switch (MERS) to serve as a voltage controller in SEIG powered DGSs. In addition, many small scale SEIGs, instead of a single large one, are applied and devoted to promote the generation efficiency. To begin with, an expandable mathematic model based on a
d

q
equivalent circuit is created for parallel SEIGs. The control method of the MERS is further improved with the objective of broadening its operating range and restraining current harmonics by parameter optimization. A hybrid control strategy is developed by taking both of the standalone and gridconnected modes into consideration. Then simulation and experiments are carried out in the case of single and double SEIG(s) generation. Finally, the measurement results verify that the proposed DGS with SVCMERS achieves a better stability and higher feasibility. The major advantages of the mentioned variable reactive power supplier, when compared to the STATCOM, include the adoption of a small DC capacitor, line frequency switching, simple control and less loss.
I. INTRODUCTION
With everincreasing concerns over environmental pollution and energy shortages, the development of renewable energy resources have attracted wide interest due to its advantages of being abundant in nature and nearly pollutant free. Small scale wind turbines and hydro turbines are widely installed in distributed generation system (DGS) for remote or isolated areas without grid coverage. The selfexcited induction generator (SEIG) is usually applied in DGSs due to its reduced installation cost, lower maintenance requirements, absence of a power supply for excitation and natural protection against system overcurrent
[1]
. However, when compared with a synchronous generator, the voltage instability and poor efficiency of a SEIG limit its application a lot. In order to steady the voltage, fixedcapacitors (FCs) and dynamic reactive power compensators (DRPCs) are needed for building noload voltage and for compensating reactive power when a load is fluctuating. Moreover, a feasible approach for promoting generation efficiency is to keep each SEIG outputting its rated power.
Some DRPCs, like thyristor switched capacitors (TSCs), thyristor controlled reactors (TCRs) and static synchronous compensators (STATCOMs), have already been designed to regulate the terminal voltage of induction generators. In
[2]
and
[3]
, the TSC and TCR are applied in a SEIG power system and a proportional integral (PI) closedloop feedback is used to realize reference voltage tracking. These articles provide a valuable method for line frequency control. However, triple sized reactors and big capacitors are employed and unacceptable current harmonics are produced at the same time. The STATCOM, as a popular DRPC, is capable of generating inductive or capacitive reactive power. Even though it has a low current distortion rate and a good dynamic response characteristic, it suffers from high frequency switching and a complicated space vector pulse width modulation (SVPWM) control
[4]

[6]
. Thyristor controlled series capacitors (TCSC)
[7]
, static synchronous series compensator (SSSC)
[8]
and unified power flow controller (UPFC)
[9]
have also succeeded in continuous reactive power regulation or power factor correction (PFC). However, they are actually compounded by TSCs or TCRs or STATCOMs and are unfit for low power applications in view of their installation cost.
In consideration of the SEIG, two kinds of mathematic models have been concluded from a lot of articles published in recent years
[10]

[14]
. For 3
Ф
SEIGs, every phase is simplified as a voltage source, a series resistance and a series inductance, and this is named the RX model
[15]
. The RX model is convenient for steady state analysis but it cannot reflect the dynamic characteristic and it increases the complicacy of modeling in parallel SEIG systems. By considering the variation of the magnetizing inductance caused by loads perturbations, state space modeling is studied in reference
[16]
. An open matrix model, which is fit for any number of parallel SEIGs, is presented in
[17]
. This is devoted to an analysis of the steady state and transient performance. With the application of the synchronous rotating coordinate conversion (SRCC), the active power and reactive power current component are easy to work out.
This paper proposes a hybrid renewable energy DGS, which mainly utilizes smallscale wind and water power generation, employs parallel connected SEIGs for energy transformation and a novel SVC device called a magnetic energy recovery switch (MERS) for voltage regulation. First, an expandable model is set up for
n
parallel SEIGs based on the SRCC. Then, the operating principle is introduced for the MERS. In order to broaden its operating range and to restrain current harmonics, an improved control method is studied by parameter optimization. The control strategy, when the DGS stands alone or is connected to a grid, is developed. Simulations and experiments are carried out to validate the performance of the MERS as a voltage regulator. The obtained experimental results support that the proposed DGS with a SVCMERS has better stability and a higher feasibility.
II. MATHEMATIC MODEL OF PARALLEL SEIGS
Since the integration of many smallscale SEIGs is more flexible and efficient under load changes,
n
different SEIGs are applied here and the
d

q
equivalent circuit of the proposed DGS is shown in
Fig. 1
. The MERS, denoted by
C
_{MERS}
, works as a controllable capacitance. The FC, denoted by
C
_{FC}
, is used to build the terminal voltage when the SEIG is not loaded. The variable load of the DGS is simplified as an inductance (
L
) and a resistance (
R
). The parameters are introduced for the SEIG in
Table І
in which
j
is the sequence number.
Dq axis equivalent circuit of the proposed parallel SEIGs using MERS to regulate terminal voltage
VARIABLE DESCRIPTION FOR SEIG
VARIABLE DESCRIPTION FOR SEIG
Based on the state space model and the open matrix model which are mentioned above, a kind of state space matrix is proposed to model parallel SEIGs as shown by equation (1). If
j
can be 1, 2, to n, the matrix of variables are defined as follows:
Coefficient matrixes, such as [
G_{j}
], [
Z_{j}
] and [
B_{j}
], are expressed in Appendix. For each SEIG, the relationship between
L_{mj}
and
I_{mj}
can be represented by the following equation:
Where
a_{j}
_{0}
~
a_{j}
_{4}
are constants. However, they are not fixed in different SEIGs. This model is used for simulation in chapter V.
III. MAGNETIC ENERGY RECOVERY SWITCH
In view of current bidirectional control, the fullbridge type MERS is adopted in this paper. Paper
[18]
points out that the equivalent capacitance can theoretically be regulated from 0 to infinity by changing the firing angle. It has been employed to correct the power factor or to stabilize terminal voltage of the inductions generator
[19]

[22]
. However, some shortcomings have gradually emerged such as an incomplete mathematical model, unacceptable harmonics, hard to realize accuracy control and a limited operating range. This section aims to rebuild the mathematic model and to broaden the regulating range.
 A. Principle and Mathematic Model
A schematic of the single phase MERS is shown on the lefttop corner of
Fig. 2
. It consists of four insulatedgate bipolar transistors (IGBTs) (G
_{1}
~G
_{4}
), a small DC capacitor (
C
_{dc}
) and a reactor (
L
_{lim}
.). In order to prevent the capacitor from being shorted, no more than two IGBTs can be opened at the same time. With the phase delay control, the parallel MERS works as a controllable capacitor and its equivalent circuit is also presented below the schematic in
Fig. 2
.
C
_{MERS}
is the equivalent capacitance.
R
_{line}
is the line resistance.
U
_{L}
,
U
_{R}
,
U
_{S}
and
U
_{MERS}
are the voltage of the power source, load, line resistance and MERS, respectively.
I
_{MERS}
represents the current. The waveforms of
U
_{S}
,
U
_{MERS}
and
I
_{M}
are plotted on the right side of
Fig. 2
in which the dotted line is the fundamental current.
Schematic and waveforms of single phase MERS.
In reference
[19]
, a MERS is used as a series compensator to regulate the power factor of an AC system and a mathematic model is built by taking the current phase for a reference. In terms of shunt compensation, this model needs to be rebuilt by taking the source voltage phase for a reference instead of the current phase. In addition, the traditional control always gets a high rate of current distortion and overvoltage of the DC capacitor
[19]

[22]
. These disadvantages lead its regulating range being limited a lot.
Taking the phase of
U
_{S}
as a reference, G
_{1}
and G
_{3}
are turned on with an alpha (
α
) delayed. Then,
C
_{dc}
starts to be charged. When
V
_{C}
is less than
V
_{Cmin}
, G
_{3}
is turned off but G
_{1}
is kept open until the
U
_{S}
phase comes to
π
+
α
. When G
_{2}
and G
_{4}
are turned on, the next half cycle can carry out with a similar process except that the current flow is reversed.
Since its waveform is obviously symmetrical to the vertical axis in
Fig. 2
, the MERS current can be defined as equation (4) by the fast Fourier transform (FFT).
Where
I
_{RMSn}
is the RMS value of the
n
^{th}
harmonic.
ω
= 2π
f
,
f
is the system frequency. The MERS voltage is formulated as:
Where
θ
=
ωt
,
X
_{Cdc}
= 1/(
ωC
_{dc}
), equation (4) can be unfolded and expressed as equation (5) by using the FFT again.
In addition, the RMS value of the fundamental voltage can be described as:
Finally, the equivalent capacitive reactance is obtained as:
 B. Variable Reactive Power Characteristic
In view of the negligible line resistance, the reactive power of the MERS branch is denoted by:
Where
E
is the RMS value of the source voltage,
X
_{MERS}
=
k
_{x}
‧
X
_{C}
and
X
_{L}
=
k
_{0}
‧
X
_{C}
,
k
_{0}
is a constant, and
k
_{x}
is a variable decided by
α
and
V
_{Cmin}
as deduced in equation (7).
F
is defined as a function with double variables,
Q
_{0}
=
ω
C
E
^{2}
.
The complexity of
X
_{MERS}
brings great difficulty in terms of ascertaining the function
F
. To study the characteristic of the output reactive power, a single phase MERS has been modeled in PSIM with the parameters shown in
Table II
.
SINGLE PHASE MERS SPECIFICATIONS
SINGLE PHASE MERS SPECIFICATIONS
An array of simulation results, by means of regulating
α
and
V
_{Cmin}
continuously and separately, has been measured. It has been traced out to compare with the theoretical curves obtained from equation (7) and (8) a shown in
Fig. 3
. The features of
Q
_{out}
between
α
and
V
_{Cmin}
include: (i) While
V
_{Cmin}
is equal to zero,
Q
_{out}
keeps growing exponentially with a rising
α
. (ii) While
α
is fixed as a constant,
Q
_{out}
decreases linearly as
V
_{Cmin}
increases. With the utilization of the linear fitting method, the proposed function
F
can be described as:
In which
K
_{α}
,
T
_{x}
and
K
_{V}
can be figured out by using the least square method (LSM).
Equivalent capacitance feature of single phase MERS.
 C. Control Parameters Optimization
Although the scope of the variable capacitance is theoretically unlimited, it only operates in a small range on account of the unavoidable ascending current harmonics in practical applications. Nevertheless, optimizing the selection of the control parameters may make it possible to reduce the production of current harmonics and it is a unique method for widening the adjusting range of a MERS to some extent. Another challenge when charging current, is the voltage of the DC capacitor which becomes too high. This is caused by the increasing to withstand itself and the IGBTs. Moreover, the current limiting reactor has a significant impact on the current quality and compensation range of the capacitive reactive power. Two solutions, regarding all of these problems, have already been enforced and they have achieved a distinct effect in terms of improving the performance of a MERS. They are introduced in detail below:
1) Make the Match of the Control Parameters
 (
α
,
V
_{Cmin}
): In reference
[21]
and
[22]
, the expected
Q
_{out}
is carried out through altering
α
or
V
_{Cmin}
asynchronously. This implies that
V
_{Cmin}
is forced to zero when
α
is adjustable and vice versa. This simple manner has disadvantages such as: (i) it is not possible to restrain current harmonics once the aim of
Q
_{out}
is ascertained. (ii) The maximum
α
, as well as the largest
Q
_{out}
, is mainly confined due to the finite ability of capacitive pressure. (iii) The growth rate of the current total harmonic distortion (THD) appears too fast to be accepted as introduced in reference
[15]
.
In consideration of the rebuilt MERS voltage equation (6), the combination control method upon
α
and
V
_{Cmin}
is probably effective in perfecting the system performance. Due to an increase of
α
or
V
_{Cmin}
, which leads to an entirely opposite effect on the reactive power production, there are at least two ways to get the expected
Q
_{out}
. In other words, any
Q
_{out}
can be acquired either by regulating
V
_{Cmin}
, by adjusting
α
, or by changing them both at the same time. There are many operating points on the equivalent power line (EPL) which get the same reactive power. Thus, one can be selected to suppress current harmonic or to avoid DCcapacitor overvoltage or breakdown. To realize the same reactive power output, two points on an EPL (5.31kVar) are applied and the characteristics via comparing the MERS voltage and current have been displayed in
Fig. 4
. The MERS peak voltage of point (25
^{0}
, 0V) is higher than that of point (28
^{0}
, 58V). The amplitude of the right side current appears lower than that of the left current. However, the triple harmonic is apparently reduced at point
V
_{Cmin}
≠ 0. This proves that there really exists an optimum point, by combining
α
and
V
_{Cmin}
, on an EPL to meet the special requirements for improving MERS performance.
Comparison of MERS peak voltage and current harmonics at different operation points.
2) Choose a Suitable Reactor:
Generally speaking, a reactor is devoted to slow down the growth of the charging/discharging current for a parallel MERS. A large inductive reactance would neutralize the emitted capacitive reactive power which will in turn narrow the compensation range. Therefore, the proportionality coefficient
k
_{0}
mentioned in equation (8) should be prudently selected.
Table III
, in which the 3
^{rd}
harmonic is excluded from the current, reflects the relationship between
k
_{0}
,
THD
_{i}
and
Q
_{out}
by simulation. Both the
THD
_{i}
and
Q
_{out}
gradually decline at any of the same working points when
k
_{0}
keeps ascending. Regardless of how the reactor changes, the
THD
_{i}
increases alone with
α
or
V
_{Cmin}
becomes larger and larger. In conclusion, the coefficient
k
_{0}
is determined by the maximum THD of the MERS current. Based on the above analysis, multiple purposes ought to be satisfied for the control parameters optimization and are stated as follows:
Where
Q
_{(}
_{α}
_{, Vcmin )}
means the generated reactive power at any working point. The first formula is elicited from equation (9). The third formula limits any dc capacitor voltage not above
V
_{Cmax}
. The last one implies that current distortion rate must be lower than the acceptable
THD
_{max}
.
CHARACTERISTICS OF DIFFERENT REACTORS EMPLOYED
CHARACTERISTICS OF DIFFERENT REACTORS EMPLOYED
3) Search for the Most Suitable α and V_{Cmin}
: With the application of synchronous control, response speed of a MERS as a DRPC is hard to accelerate unless the precision of every controlling cycle is enhanced. In view of the
α
changed in discrete from, the output reactive power must be discontinuous. Therefore, the expected
Q
_{out}
can be divided into two parts as follows:
Where
Q
_{α}
is generated by α. Δ
Q
is an error adjusted by
V
_{Cmin}
.
Actually, α and
V
_{Cmin}
are complementary in terms of producing continuous reactive power as shown in
Fig. 5
. The dotted line is a cluster of EPL. In the process of the reactive power changed from
Q
_{j}
_{1}
to
Q
_{j}
, α is first increased to reach the EPL
Q
_{α}
, and then
V
_{Cmin}
is regulated to reach the target value,
Q
_{out}
. The third step is searching for the most suitable point based on equation (9). In
Fig. 5
, the bold solid lines mean the practical searching path for obtaining the most suitable working point. The special
V
_{Cmin}
can be calculated by
V
_{0}
+
K
Δ
Q
, in which
V
_{0}
and
K
are constants which are worked out by the linear fitting method (LFM).
Approach applied to search the most suitable EPL point
4) Improved Performance as a Shunt DRPC:
With the proposed parameters optimization control, a lot of working points for the MERS are increased. When compared to reference
[21]
, every
Q
_{out}
corresponds to an EPL line but not a single point. Furthermore, the regulating range of
X
_{MERS}
is broadened and satisfies the conditions of the
THD
_{i}
and
V
_{Cpeak}
limitations of grid. The curves, as measured by simulation, reveal the effect as shown in
Fig. 6
(a) and
Fig. 6
(b). The dotted lines, with the application of the control method mentioned in reference
[21]
, are measured for comparison. When the
THD
_{i}
is lower than 6.0% and
V
_{Cpeak}
does not exceed double
V
_{C0}
, the maximum
X
_{MERS}
rises from 1.2 p.u to 1.4 p.u, and the minimum
X
_{MERS}
declines from 0.7 p.u to 0.35 p.u.
V
_{C0}
is the capacitor peak voltage of the working point (0
^{0}
, 0V).
Comparison of regulating range of X_{MERS}.
 D. Development of ThreePhase MERSs
For a 3
Ф
system, three MERSs are used, as shown in
Fig. 7
, and connected in delta formation for eliminating the triple current harmonics. The load is equivalent to a changeable resistance and inductance. Power can also flow into the AC grid by operating switch ‘S’ in case it is needed. Each MERS is controlled independently and IGBTs in the same position but in different MERSs are switched by threephase symmetrical signals. Therefore, only four control signals needs to be supplied even though double the number of switches is used when compared to STATCOMs. In addition, the MERS works in the line frequency and generates less switching loss. With the help of FCs, the 3
Ф
MERSs can supply a large range of reactive power for SEIG power systems. A kind of multiple SEIGs power system is going to be studied and tested by simulation and experiments in section V.
Schematic of SEIGbased power system using a three phase MERSs as a voltage or reactive power regulator.
IV. SYSTEM CONTROL
In order to achieve voltage phase detection and magnitude measurement, the phase lock loop (PLL) technique is studied in this section. Then, a
d

q
model of the three phase deltaconnected MERSs is created and a uniform control strategy is proposed for voltage regulation under the state of standalone operation and power factor correction in the case of connection to the grid.
 A. Phase Lock Loop Technique
Assume that the input three phase voltages are three cosine functions with the same amplitude (
U
_{m}
), but with phases delayed 120
^{0}
in sequence. Therefore, a new equation is obtained as:
Where
θ
is the synchronous phase of the voltage, and
θ
^{*}
is the actual phase. When the error between
θ
and
θ
^{*}
becomes very small,
v
_{d}
and
v
_{q}
tend to be
U
_{m}
and
U
_{m}
‧ (
θ

θ
^{*}
), respectively. Therefore, a closed loop control is proposed to gain a precise voltage phase and amplitude as shown in
Fig. 8
.
Schematic of applied closed loop control.
The transfer function of this control loop is revealed as
H
(s). Bode diagrams of
H
(s) are depicted in
Fig. 9
according to different control parameters,
k
_{P}
and
k
_{I}
. The performance of ζ=0.707 is more stable and is selected here.
Where,
ζ
is defined as follows:
Bode diagrams of H(s) for different values of ζ.
Furthermore, the three phase currents are also assumed to be cosine functions, but with the same magnitude (
I
_{m}
). The
d

q
current components are described as:
Where
I
_{AP}
and
I
_{RP}
refer to the active power current component and the reactive power current component, respectively.
 B. Control Strategy
With the use of the mentioned PLL, the
d

q
model of a 3
Ф
deltaconnected MERSs can be described as:
Where
E
_{d}
,
E
_{q}
,
U
_{MERSd}
,
U
_{MERSq}
,
I
_{MERSd}
and
I
_{MERSq}
refer to the grid voltage, MERS voltage and MERS current, respectively. Only the lower formula of equation (15) is studied on account of the negligible active power current component.
To realize stability regulation, a negative feedback function is introduced and defined as:
Where
K
_{MERS}
and
T
_{MERS}
are the proportionality coefficient and the time constant, respectively. The working state of this DGS can be described by a switch function as:
Then, the reference current (
I
^{*}
_{MERSq}
) can be expressed as:
Where
U
_{REF}
means the reference phase voltage of the DGS, and
I
_{FCq}
,
I
_{Lq}
and
I
_{siq }
are the reactive power current components of each of these branches.
K
_{u}
and
T
_{u}
are the proportionality coefficient and time constant of this voltage control loop. In addition,
Q
_{out}
is obtained on the basis of equations (8) ~ (10) as:
From equation (8), the primary operating point (
α
, 0) is ascertained by:
Then,
V
_{Cmin}
can be worked out by equation (9).
The proposed control strategy is shown in
Fig. 10
. The voltage or current signals are detected from the DGS at first and transformed to
d

q
axis signals using the SRCC. The reference current,
I
^{*}
_{MERSq}
, is determined by two different operating modes. Then, equation (16) and (18) are applied to get the control variable,
U
_{MERSq}
. The expected
Q
_{out}
can be achieved with equation (19) and used for EPL confirmation. Lastly, the control parameters are optimized and the control signals are generated to drive the 3
Φ
MERSs to provide DRPC for the DGS.
Schematic of the proposed control strategy.
V. SIMULATION AND EXPERIMENTS VERIFICATION
A smallscale DGS with two SEIGs has been developed as well as an equivalent simulation model based on equation (1) and the same parameters used in PSIM. The 3
Φ
MERSs and groups of FCs are employed to supply continuous reactive power. First, the proposed control strategy is demonstrated in a developed 3
Φ
MERSs when it connects to a power grid directly. With the novel DRPC device, the adjustment of the system voltage or reactive power are verified by both the simulation and experiment results when only one SEIG is powered and the double SEIGs are powered, respectively.
 A. Experimental System Construction
A principle diagram of the experimental DGS is shown in
Fig. 11
. A selfexcited induction motor (SEIM) is selected to simulate a wind turbine as a prime mover. SW1, SW2, SW3 and S, are used to realize the different running modes.
Diagram of the system construction of experimental DGS.
The system specifications are revealed in
Table IV
. A photograph of the designed 3
Φ
MERSs is shown in
Fig. 12
(a), and
Fig. 12
(b) shows a simulation of the wind power system. Expert III, which works as a central controller with an integrated DSP (C6713), is applied here. PEView 9 is a kind of special control soft for Expert III. The testing equipment includes a TPS 2024 (200MHz), a FLUKE 43B POWER QUALITY ANALYZER, etc. In order to achieve practical performance, the parameters of the induction motor are measured and used in the simulation model, as shown in Appendix
Table V
.
SPECIFICATIONS OF THE EXPERIMENTAL DGS
SPECIFICATIONS OF THE EXPERIMENTAL DGS
Photographs of the experimental devices.
MAIN PARAMETERS OF SEIGS
 B. Verification of the Proposed DRPC
Fig. 13
shows a flow diagram of the applied program. This part only takes the situation of S=0 into account to show that the MERSs can produce a continuously variable reactive power. Some necessary constants are ascertained separately:
THD
_{max}
=6.0%,
V
_{Cpeak}
=1000V,
Q
_{0}
=2.268kvar, ζ=0.707,
K
_{MERS}
=1.15,
T
_{MERS}
=0.0025, and
E
_{q}
=0. Then, experimental results are obtained as shown in the following pictures.
Schematic of the program running in PEView 9
1) Performance of the Optimized Operating Point:
With the parameter optimization control, the MERS current THD and
V
_{Cpeak}
are adjusted by
V
_{Cmin}
.
Fig. 14
and
Fig. 15
reveal the simulation and experimental results, respectively. In
Fig. 14
(a), both of the two points can generate the same reactive power (EPL=12.45kvar). However, the right figure shows the lower peakpeak MERS voltage and current distortion rates achieved at point (10
^{0}
, 15V).
Fig. 14
(b) demonstrates this feature when the EPL is equal 16.12kvar.
Fig. 15
(a) and
Fig. 15
(b) show the experimental results which show that the proposed method is effective.
Simulation of parameters optimization control.
Experiment of control parameters optimization.
In addition, different operating points and the corresponding
Q
_{out}
are applied to calculate some of the coefficients mentioned above using LFM and
K
_{α}
=1.1,
T
_{x}
=0.033,
K
_{v}
=0.002, V
_{0}
=0.00, and
K
=0.05.
2) Reactive Power Compensation Range Extended:
When the conditions of equation (10) are satisfied, the practical range of the reactive power compensation and the operating point are considered and experimented on.
Fig. 16
(a) shows the
THD
_{i}
and the capacitor peak voltage when the output reactive power rises and
Fig. 16
(b) gives its permit working point. The squares connected by solid lines are the experimental results with the proposed parameter optimization control. The squares connected by dotted lines are the experimental results using the control method before improvement. The filled squares reflect the peak voltage of the capacitor and unfilled ones refer to the MERS current THD.
Steady state characteristics of MERSs.
The capacity of the reactive power compensation can be obtained and it can be regulated from 7.5kVar to 22.5kVar when the
THD
_{i}
<6% and the
V
_{Cpeak}
<1000V as shown in
Fig. 16
(a).
In
Fig. 16
(b), the operating points of the improved MERS are broadened to the large area which is enclosed by four bold lines. In reference
[21]
, the working point is only permitted on the α axis or the
V
_{Cmin}
axis, and the range of the dynamic reactive power is just regulated from 8.0kVar to 16.0kVar. Therefore, both the minimum and maximum
Q
_{out}
are extended in this work.
 C. Voltage Regulation for the SEIGs Powered DGS
In this part, the proposed control strategy is verified with the switches S, SW1 and SW3 closed as shown in
Fig. 11
. The MERSs and the FCs just work as power factor correctors (PFC) for the grid when load is changing. Then, S is open and the voltage of the SEIGs is regulated by the MERSs. Lastly, SW2 is also closed and the experimental results are separately measured when the load goes up or down.
1) Power Factor Correction of the Grid Connected SEIG:
When a SEIG connects to a grid, the output voltage is always kept at a rated value since it is fixed by the grid voltage. The most important feature is the transient and steady state performance in terms of power factor correction when the MERSs are applied. With the load increasing or declining, the dynamic and stable characteristics are verified by simulation and experiment studies.
In order to achieve the best performance in terms of power factor regulation, the proportion of
K
_{MERS}
and
T
_{MERS}
needs to keep at 1:1000, which is verified by the simulation. Its effect is shown in
Fig. 17
(a) and
Fig. 17
(b) under the conditions of an increasing or decreasing load.
Fig. 18
(a) and
Fig. 18
(b) show the dynamic process in the experiments.
Transient characteristics when the proportion between K_{MERS} and T_{MERS} is 0.001.
Transient characteristic when K_{MERS} =1.22 and T_{MERS} =1350.0 in experiment.
By continuously increasing the load, the power factor of the grid is measured and as shown in
Fig. 19
. The load reactive power is dynamically compensated by the MERSs and FCs as revealed by the solid line of connected circles. The squares connected by solid line or dotted line, show the steady state process of the grid PFC in the simulation and experiment, respectively. Obviously, the power factor can always maintain a unit because the MERSs produce equivalent reactive power to make up for the SEIG leaded by increasing the load. When the output reactive power is lower than 7.5kVar or larger than 22.5kVar, the current THD is measured and shown in
Fig. 20
(a) and
Fig. 20
(b).
Steady state performance on power factor regulation.
Current THD for the boundary operating point.
2) Voltage Regulation for a Single SEIG:
Fig. 21
(a) and
Fig. 21
(b) show the processes of noload voltage building with FCs excited and with MERSs. The simulation model shown in
Fig. 7
has been built in PSIM. The parameters of this model are the same as those used in the experimental SEIGs as shown in
Table V
in the Appendix.
The process of noLoad voltage building by experiment.
The first step is to ascertain a couple of
K
_{U}
and
T
_{U}
to get the best transient characteristic. When
K
_{U}
=1.0 and
T
_{U}
=1000.0, the simulation results, including the waveforms of three phase voltage, are described in
Fig. 22
. The bold line means the RMS value of the grid voltage. Then, a group of special
K
_{U}
and
T
_{U}
is estimated and used in the experiment on the basis of the simulation results.
Fig. 23
shows the experimental results when
K
_{U}
=3.88 and
T
_{U}
=3550.0.
SEIG’s lineline voltage regulation by Simulation.
Experimental results of SEIG voltage adjusting when K_{U} =3.88 and T_{U} =3550.0.
3) Voltage Regulation for Double SEIGs:
With the parameters
K
_{U}
and
T
_{U}
tested above, the dynamic process of voltage regulation at the moment when a 4 kW SEIG is inserted in or cut off is shown in
Fig. 24
.
Voltage regulating at the moment of 4kW SEIG inserted in or cut off.
Finally, the voltage regulation of the developed DGS is realized and the curves, as shown in
Fig. 25
, reflect the characteristic of its steadystate performance. Obviously, the grid voltage is dramatically reduced, as revealed by the dotted line with filled squares, unless the MERSs are installed. The solid line with unfilled circles shows that the output of the MERSs reactive power increases continuously with the rising load power. The unfilled squares which are connected by a solid line and a dotted line, verify that the voltage stability of the DGS can be maintained by the MERSs + FCs. In conclusion, the MERS works as a DRPC in SEIGs powered wind systems and it is verified by both simulation and experimental results.
Stable performance of the developed DGS voltage.
VI. CONCLUSION AND FUTURE WORK
A simple fullbridge MERS which is especially applicable to smallscale DGSs has been studied in this article. A theoretical analysis and experiments are devoted to validate it as a DRPC and to illustrate its feasibility for voltage or reactive power regulation in SEIGs power systems. Both the steady state and transient state experimental results support the conclusion that the MERS is a valuable and reliable reactive power controller. When compared to STATCOMs, the major advantages of the MERS include the adoption of a small DC capacitor, line frequency switching, simple control, and reduced losses. The MERS can also realize reactive power control in microgrids. This will be investigated at some point in the future.
Acknowledgements
This work is supported by Youth Fund of National Natural Science Foundation in China (Grant no.6110418).
BIO
Yewen Wei was born in Hunan province, China. He received his B.S. degree in Measurement and Control Technology for Instrument from Xiangtan University, Xiangtan, China, in 2008. Since 2009, he has been working towards his Ph.D. degree in the School of Electric Power, South China University of Technology, Guangzhou, China. His current research interests include power electronic converters, the application of power electronics in renewable energy systems, reactive power control, harmonics, and power quality compensation systems such as SVC, Magnetic Energy Recovery Switch (MERS), and FACTS devices.
Longyun Kang was born in Jilin, China, in 1961. He received his B.S., degree in Physics from Yanbian University, Yanji, China, in 1982, and his M.S. and Ph.D. degrees in Electrical Engineering from the Department of Engineering, Kyoto University, Kyoto, Japan, in 1996 and 1999, respectively. From 1999 to 2001, he was a Researcher with the Department of Engineering, Tokyo Institute of Technology, Tokyo, Japan. From 2001 to 2006, he was an Associate Professor with the Institute of Mechanical Engineering, Xi’an Jiaotong University, Xi’an, China. Since 2006, he has been with the School of Electric Power, South China University of Technology, Guangzhou, China, where he is presently a Professor, a supervisor of Ph.D. students, and a director of the Guangdong Key Laboratory of Clean Energy Technology. His current research interests include renewable energy and electric vehicles, including wind energy, solar energy conversion, hybrid energy systems and hybriddrive technology for electric vehicles.
Zhizhen Huang was born in Anhui, China, in 1990. He received his B.S. degree from Xiangtan University, Xiangtan, China, in 2013. Since 2013, he has been working towards his M.S. degree in the School of Electric Power, South China University of Technology, Guangzhou, China. His current research interests include power electronic converters, and MERS based wireless charging technology for vehicles.
Zhen Li was born in Hunan province, China, in 1992. He received his B.S. degree from the China University of Mining and Technology, Xuzhou, China, in 2014. Since 2014, he has been working towards his M.S. degree in the School of Electrical Power, South China University of Technology, Guangzhou, China. His current research interests include power electronic converters, and power quality compensation systems such as SVC and Magnetic Energy Recovery Switch (MERS).
Miao miao Cheng was born in China, in 1982. She received her M.S. degree from Xi'an Jiaotong University, Xi’an, China, in 2006, and her Ph.D. degree from the Tokyo Institute of Technology, Tokyo, Japan, in 2009, where she worked as a Postdoctor for the following three years. Since 2012, she has been working as an Assistant Professor at Hunan University, Changsha, China. Her current research interests include motor control, energy storage and power electronics.
Touti E.
,
Pusca R.
,
Manata J. P.
,
Brudny J. F.
,
Chaari A.
2014
“On the use of a dimmer for a robust frequency control of a selfexcited threephase induction wind generator,”
Journal of Power Electronics
14
(3)
580 
591
DOI : 10.6113/JPE.2014.14.3.580
Ahmed T.
,
Noro O.
,
Hiraki E.
,
Nakaoka M.
2004
“Terminal voltage regulation characteristics by static VAR compensator for a threephase selfexcited induction generator,”
IEEE Trans. Ind. App.
40
(4)
978 
988
DOI : 10.1109/TIA.2004.830783
Spasojević L.
,
Papič I.
,
Blažič B.
2014
“Development of a control algorithm for a static VAR compensator used in industrial networks,”
Journal of Power Electronics
14
(4)
754 
763
DOI : 10.6113/JPE.2014.14.4.754
Singh B.
,
Murthy S S.
,
Gupta S.
2006
“STATCOMbased voltage regulator for selfexcited induction generator feeding nonlinear loads,”
IEEE Trans. Ind. App.
53
(5)
1437 
1452
Barrado J.A.
,
Griñó R.
,
ValdBlavi H.
2010
“Powerquality improvement of a standalone induction generator using a STATCOM with battery energy storage system,”
IEEE Trans. Power Del.
25
(4)
734 
2741
DOI : 10.1109/TPWRD.2010.2051565
Ahmed T.
,
Nishida K.
,
Nakaoka M.
2007
“A novel stand alone induction generator system for ac and DC power applications,”
IEEE Trans. Ind. App.
43
(6)
1465 
1474
DOI : 10.1109/TIA.2007.908210
Tiwari P. K.
,
Sood Y R.
2013
“An efficient approach for optimal allocation and parameters determination of TCSC with investment cost recovery under competitive power market,”
IEEE Trans. Power System
28
(3)
2475 
2484
DOI : 10.1109/TPWRS.2013.2243848
ElMoursi M. S.
,
Sharaf A. M.
2005
“Novel controllers for the 48pulse VSC STATCOM and SSSC for voltage regulation and reactive power compensation,”
IEEE Trans. Power Syst.
20
(4)
1985 
1997
DOI : 10.1109/TPWRS.2005.856996
An S.
,
Condren J.
,
Gedra T.W.
2007
“An ideal transformer UPFC model, OPF firstorder sensitivities, and application to scree for optimal UPFC locations,”
IEEE Trans. Power System
22
(1)
68 
75
DOI : 10.1109/TPWRS.2006.889141
Kasal G. K.
,
Singh B.
2011
“Voltage and frequency controllers for an asynchronous generatorbased isolated wind energy conversion system,”
IEEE Trans. Energy Convers.
26
(2)
402 
416
DOI : 10.1109/TEC.2010.2102029
Kumar S.
,
Narayan R.
2012
“Effect of capacitive VAR on performance of threephase selfexcited induction generator,”
International Journal of Emerging Technology and Advanced Engineering
12
(2)
253 
258
Bansal R. C.
2005
“Threephase selfexcited induction generators: An Overview,”
IEEE Trans. Energy Convers.
20
(2)
292 
299
DOI : 10.1109/TEC.2004.842395
Boora S.
2010
“Analysis of selfexcited induction generator under balanced or unbalanced conditions,”
International Journal on Electrical and Power Engineering
1
(3)
59 
63
Seyoum D.
,
Grantham C.
,
Rahman M. F.
2003
“The dynamic characteristics of an isolated selfexcited induction generator driven by a wind turbine,”
IEEE Trans. Ind. Appl.
39
(4)
936 
944
DOI : 10.1109/TIA.2003.813738
Cidrás J.
,
Feijóo A. E.
2002
“A linear dynamic model for asynchronous wind turbines with mechanical fluctuations,”
IEEE Trans. Power Syst.
17
(3)
681 
687
DOI : 10.1109/TPWRS.2002.800912
Kumar G. S.
,
Kishore A.
“Dynamic analysis and control of output voltage of a wind turbine driven isolated induction generator,”
in Proc. ICIT
2006
494 
499
Farret F. A.
,
Palle B.
,
simoes M. G.
“State Space Modeling of Parallel SelfExcited Induction Generator for Wind Farm Simulation,”
in Proc. IAS
2004
2801 
2807
Isobe T.
,
Wiik J. A.
,
Kitahara T.
,
Kato S.
,
Inoue K.
“Control of series compensated induction motor using MERS,”
in Proc. PEA&EC
2007
1 
10
Wiik J. A.
,
Fonstelien O. J.
,
Shimada R.
“A MERS type Series FACTS Controller for Low Voltage Ride Through of Induction Generators in Wind Farms,”
in Proc. PEA&EC
2009
1 
10
Kawaguchi T.
,
Sakazaki T.
,
Isobe T.
2013
“Offshore wind farm conﬁguration using diode rectiﬁer with MERS in current link topology,”
IEEE Trans. Ind. Electron.
60
(7)
2930 
2937
DOI : 10.1109/TIE.2012.2217716
Cheng M. M.
,
Shiojima D.
,
Isobe T.
,
Shimada R.
“Voltage control of induction generator powered distributed system using a new reactive power compensator SVCMERS,”
in Proc. PEMCC
2013
71 
78
Wijaya F. D.
,
Isobe T.
,
Usuki K.
,
Wiik J. A.
,
Shimada R.
“a new automatic voltage regulator of selfexcited induction generator using SVC magnetic energy recovery switch (MERS),”
in Proc. IEPESC
2008
697 
703