This paper presents a novel approach for achieving both a tight DC voltage regulation and a power factor control by applying the Reaching Law Sliding Mode Control (RLSMC) and the conventional Sliding Mode Control (SMC). Applying these strategies on a matrix rectifier (MR) can achieve a unity grid side power factor when the DC load changes widely and it can provide a ripplefree output voltage that is easily affected by distortions of the threephase ac voltage supply. Furthermore, by employing the reaching law on the SMC can solve the chatting problem of the sliding motion. Comparative Matlab simulations and experimental verifications for these strategies have been presented and discussed in this paper. The results show that by applying the SMC and RLSMC on a MR can achieve a unity grid side power factor and a regulated ripplefree DC output.
I. INTRODUCTION
The conventional rectification, uncontrolled rectifiers and phase controlled rectifiers, may bring about a large number of harmonics, which have become the main ‘pollution’ of the grid. In addition, such ‘pollution’ makes other electric devices work improperly. Matrix Converters (MC) can be traced to the late 1970s
[1]
. The three phase to two phase ACDC MC, i.e. matrix rectifier, can be deduced from the conventional three phase to three phase MC. The MR has several advantages: (1) operation in all four quadrants; (2) no need for a large value DC capacitor on the output side; (3) easy safe current commutation; (4) intrinsic buck converter and (5) power factor regulation
[2]

[3]
. Because of the advantages mentioned above, the MR has become a strong candidate for potential applications in plugin automotive, aircraft, DC motor drives, etc.
[4]
. Moreover, the MR is the first stage of a twostage matrix converter, and it can be used in the field of wind power generation and has wide application prospects and potential economic value.
Commonly there are several modulation algorithms for the MR, and they are the Alesina–Venturini optimum (AV optimum) method
[5]
, the Roy method
[6]
and the Space Vector Modulation (SVM)
[7]
. However, these algorithms are openloop control. When applying these modulation algorithms on a MR, the performance of the DC output voltage becomes bad, since such modulation algorithms cannot resist disturbances.
An input filter is necessary for a sinusoidal input current because it can filter the harmonic components of the input current. However, the input filter’s characteristics make the input power factor (IPF) remain high in the presence of high output loads, and the IPF degrades significantly under lightload conditions
[8]
. What is more, due to the presence of a grid side unbalanced input voltage and voltage disturbances, a MR cannot provide a ripplefree DC output voltage, and periodic oscillations appear.
In order to mitigate the fluctuations of the DC output voltage cause by the grid side unbalanced input voltage and to improve the IPF, scholars have carried out a lot of research in these fields. In
[9]
, the static and power characteristics of the MR have been given. This has been a crucial support in terms of further research on control strategies. In
[10]
, the designs of a feedforward control strategy was proposed to guarantee the stability of the DC output voltages. In
[11]
, the designs of an openedloop control, and in
[12]

[13]
, the designs of closedloop control strategies were proposed to compensate the power factor. As a result, applying feedforward control strategies to improve the performance of DC output voltage regulation can restrain disturbances of the unbalanced threephase input to a certain extent. However, the performance is not obvious. In
[14]
, a newdirect SVM based on an API controller was shown, and its correctness and effectiveness had been verified. However, PI control is a linear control, and its dynamic properties is poor. In
[15]
, in order to improve dynamic performance, the conventional SMC was applied to compensate the input power factor and the DC output. A PI controller and a SMC controller were employed separately in the prototype experiment. The experiment results reveal that the dynamic properties of the SMC are superior. However, the output ripple voltage and input current harmonics are much more serious than those regulated by a PI controller.
In this paper, a novel approach to achieve both a tight DC voltage and a grid side power factor is proposed. The DC output voltage is controlled by a RLSMC and the grid side power factor is compensated by the conventional SMC. Simulation and experimental results show that the proposed algorithm can fulfill the presupposed goals and alleviate chattering of the sliding mode.
II. MATHEMATICAL MODEL OF A PMSM FOR BACKEMF ESTIMATION
 A. Space Vector Modulation Method
Fig. 1
shows the topology of the MR and this topology consists of 6 parts: a threephase power source, input damping filters, a main circuit with six bidirectional switches, an output filter, and a load.
Topology of the MR.
Space vector modulation (SVM) is used to provide sinusoidal input currents for the MR. The SVM must satisfy the following two constrains simultaneously:
(1) The existence theorem of the matrix converter theories in
[16]

[17]
.
(2) The maximization of the DC output voltage to achieve full use of the input line–line voltages.
To satisfy the above constrains, only nine feasible switching states of the six bidirectional switches exist, and these nine states determine nine vectors which can be divided into two categories: active vectors (
I
_{1}

I
_{6}
) and zero vectors （
I
_{7}

I
_{9}
）.
In addition, the space vector hexagon is divided into six sectors by the active vectors, as shown in
Fig. 2
. The reference current vector
I
_{ref}
, synthesized from the two adjacent vectors
I
_{α}
and
I
_{β}
, and a zero vector, as shown in
Fig. 3
.
θ
is the angle between
I
_{ref}
and its right adjacent active vector
I
_{α}
, and
θ
∈[0,
π
/3]. The reference current vector
I
_{ref}
can be calculated as the following expression:
where
T
_{α}
(
θ
),
T
_{β}
(
θ
) and
T
_{0}
(
θ
) are the durative times of
I
_{α}
,
I
_{β}
and
I
_{0}
in one switching period
T_{S}
. The duty cycle
d
_{α}
(
θ
),
d
_{β}
(
θ
) and
d
_{0}
(
θ
) can be calculated as follows:
where
m
is the modulation index,
m
∈[0,1]. The averaged output voltage in a switching period can be calculated as:
where
V
_{im}
is the amplitude of the input phase voltage, and
φ
_{i}
is the desired input current displacement angle.
Space vector diagram of input current.
Reference inputcurrent space vector with the adjacent vectors.
 B. Input filter Analysis of the MR
The input filter, as shown in
Fig. 4
, is designed to filter the high harmonic components of the input current and to reduce the input voltage distortion supplied to the MR. In this topology, to damp the oscillations resulting from the matrix rectifier hard switching, the resistance is connected in parallel with the inductance, and this topology provides the minimization of power losses
[18]
.
The input filter of the MR.
In the
Fig. 4
,
U
_{s}
and
I
_{s}
denote the input voltage and current at the grid side, and
U
_{i}
and
I
_{i}
denotes the input voltage and current of the MR. The IPF issue is addressed only when the input filter is considered. At the grid side, let
U
_{s}
=
U
_{sm}
∠
α
and
I
_{s}
=
I
_{sm}
∠
β
. At the output side of the filter, let
I
_{in}
=
I
_{im}
∠(
α

φ
_{i}
). To analyze
Fig. 4
, it should be known that:
It should be noted that
is so small that it can be ignored. Eq. (6) can be simplified as:
According to Eq. (7) and
Fig. 5
, the grid side power factor can be calculated as follows:
where
α
and
β
are the phase angles of the input voltage and input current at the grid side, respectively,
φ
_{s}
is the grid side power factor angle, and
φ
_{i}
is the switch matrix power factor angle. From Eq.(8), the phase shift
φ
_{s}
is imposed by the MR’s input filter, and the phase shift relates to the whole circuit including the DC loads.
Vector diagram of the Eq.(7).
 C. Impact of the DC Output Voltage Caused by an Imbalanced Input Voltage
According to the symmetrical components method, an unbalanced threephase input voltage
V_{s}
can be decomposed into positive sequence component
V
_{sp}
and negative sequence component
V
_{sn}
as follows:
In Eq. (9),
V
_{PM}
and
V
_{NM}
denote the amplitudes of the positive and negative sequence input phase voltages, respectively, and
ω
is the input voltage angular frequency. Let the initial phase angles of the positive and negative sequence components become zero.
In
[19]
:
Hence, the DC output voltage can be shown as:
In Eq. (11), under the imbalanced threephase voltage input condition, the output DC voltage of the Matrix Rectifier contain a ripple wave with a frequency that is double that of the input ripple wave except for the desired DC output voltage (as shown in
Fig. 6
).
Input and Output voltage waveform of the MR.
III. SLIDING MODE CONTROL SYSTEM BASED ON THE REACHING LAW
The sliding mode control is one of the variable structure control strategies (VSC). The SMC has several advantages such as a quick response, strong robustness, and simple implementation. In addition, it is not affected by system parameter and external disturbances and it has good dynamic performance. However, the SMC makes the system switch between two kinds of states, which results in a tremendous chattering around the sliding surface when the system operates in the sliding mode. These problems have been tackled by using the Reaching Law on the SMC.
 A. Mathematical Model of the System
After tangent calculus is applied to Eq. (8), it can be shown as:
where
φ
_{s}
is the grid side power factor angle. Assuming
φ
_{s}
<
π
/6 and
φ
_{i}
<
π
/6, the output power of the MR is shown as
Therefore, Eq. (8) can be shown as:
From
Fig. 1
, according to Kirchhoff’s law, the following two equations are known:
Thus, the equation for the state of the output voltage and current can be given as:
Eqs. (5), (14) and (15) are the key factors for the proposed mathematical and control models.
 B. Design of the Slide Mode Controller
According to the analysis above, the principle diagram of the slide mode controller operational principle is shown in
Fig. 7
. The working process is as follows: the sensor detects the DC output voltage
V
_{o}
that will be compared with the desired output voltage
V
_{ref}
. This determines the error value. Then, the error value is sent to the RLSMC controller based on the reaching law. Then, the modulation index
m
will be calculated. The sensor detects the phases of the input voltages and currents at the grid side. This determines
φ
_{s}
, the power factor at the grid side. The phase difference
φ
_{s}
is sent to the PF SMC controller. Then the phase difference
φ
_{i}
is determined, and the compensation angle is calculated.
The principle diagram.
 B. DC Output Voltage Regulation by the RLSMC
The control system is designed to suppress the DC output voltage fluctuations caused by an unbalanced input threephase voltage. The controller allows the DC output voltage
V
_{o}
track the reference voltage
V
_{ref}
.
Where
e
_{v}
is the feedback voltage error value. In addition,
Thus, Eq.(16) is shown as:
The slide mode function is shown as:
where
c
_{1}
is a positive constant.
In Eq. (19), when
c
_{1}
≈
R
_{L}
C
_{O}
,
or
Thus, Eq. (19) can be simplified as:
In order to achieve a higher dynamic performance of the DC output voltage, the exponent reaching law is applied:
Where
ε
_{1}
and
k
are positive constants. From Eq. (20) and Eq. (21), it can be know that:
From Eq. (5) and Eq. (22), the modulation index
m
can be shown as:
 C. Grid Side Power Factor Compensated by the Constant Reaching Law
The control system is also designed to compensate the power factor at the grid side. This control algorithm is able to compensate the phase of the input current at the grid side by applying the SMC. Thus, it can guarantee that the phase
φ
_{s}
is zero in theory. The switching function is set as:
Where,
is the average rate of change. Meanwhile the control function is as follows:
where,
φ
_{ref}
is the precompensation angle. The phase
δ
is the disturbance resist factor and 
φ
_{i}
≤
π
/6 . To achieve the local reachability of theSMC, Eq. (25) should be satisfied as:
where
Ṡ
_{2}
is the time derivative of
S
_{2}
.
If
c
_{2}
is small enough,
S
_{2}
=
φ
_{s}
and Eq. (27) can be simplified as follows:
From Eq. （13）,
and
Because
φ
_{s}
and
φ
_{i}
are nonlinear functions,
and
are the change rates of
φ
_{s}
and
φ
_{i}
in the control cycle. As a result,
and
are also the change rates of
and
.
From Eq. (25), in the next control cycle, there is
φ
_{i}
=
φ
_{ref}

δ
. Meanwhile
= (
δ

δ
)
f_{s}
=2×10
^{4}
δ
and
= (2
δf
_{s}
 2
δf_{s}
)
f_{s}
= 4×10
^{8}
δ
, where
f
_{s}
is the switching frequency and
f
_{s}
=10
^{4}
. Substituting the abovementioned expressions of
and
into the Eq. (28), inequality (26) holds, which means that the sliding mode exists
[20]
.
From Eq. (25), in the next control cycle, there is
φ
_{i}
=
φ
_{ref}
+
δ
. Meanwhile
= (
δ
+
δ
)
f_{s}
= 2×10
^{4}
δ
and
= (2
δf_{s}
+ 2
δf_{s}
)
f_{s}
= 4×10
^{8}
δ
. Substituting
and
into Eq. (28), inequality (26) also holds, which means that the sliding mode exists.
The value range discussion of the sliding mode coefficient: if
c
_{2}
is small enough to make the equality
, then
, consequently:
From Eq. (29), the value of the sliding mode coefficient is related to the switching frequency and the control function (25).
IV. SIMULATION AND EXPERIMENTAL RESULTS
In this section, a detailed simulation is built on the MATLAB/SIMULINK platform to verify the proposed control strategy.
 A. Simulation and its Analysis
The circuit parameters listed in
Table I
are used in the smallscale simulation model. Three comparative simulations were carried out in this section: (1) balance input with different values of
c
_{1}
; (2) threephase balanced input voltage using the conventional SMC; (3) threephase balanced input voltage using the SMC based on the reaching law; and (4) threephase unbalanced input voltage using the SMC based on the reaching law. The simulation waveforms and the results of the total harmonic distortions are shown in
Fig. 9
to
Fig. 11
.
PARAMETERS OF THE SIMULATION
PARAMETERS OF THE SIMULATION
Fig. 8
shows a simulation with different values of
c
_{1}
under the balance input condition. In simulation (a)
c
_{1}
=6×10
^{4}
=R
_{L}
C
_{O}
and in simulation (b)
c
_{1}
=2×10
^{6}
_{L}
C
_{O}
. From the simulations results it can be seen that the former DC output performance is worse than latter since a larger value of
c
_{1}
makes the
c
_{1}
ė
_{v}
part become larger and deteriorates the control precision.
Experimental results when the input is balanced but with different value of c_{1}: (a) c_{1}=6×10^{4}; (b)c_{1}=2×10^{6}.
A comparative simulation study was conducted using the same circuit parameters but with different control strategies. Both the conventional SMC and the SMC based on the reaching law were used. According to
Fig. 9
and
Fig. 10
, using the two abovementioned approached can achieve two goals, ripplefree output voltage and a near unity power factor. Moreover the chattering of the waveform is acceptable. At 50 ms, the target output voltage dropped from 80 V to 50 V. The simulation results proved that the dynamic performance was excellent regardless of which control algorithm was used.
Experimental results of using conventional SMC when the input voltage is balanced: (a) simulation waveform; (b) THD analysis.
Experimental results of using SMC based on reaching law when the input voltage is balanced: (a) simulation waveform; (b) THD analysis.
According to
Fig. 11
, when the threephase input voltage is unbalanced, applying the RLSMC algorithms on the MR can provide a ripplefree output voltage.
Simulation waveform using SMC based on reaching law when the input voltage is unbalanced.
 B. Experimental Results and their Analysis
An experimental prototype was built to further illustrate and validate the proposed control approach. Three comparative experiments were carried out: (1) a threephase balanced input voltage using the conventional SMC; (2) a threephase balanced input voltage using the SMC based on the reaching law; and (3) a threephase unbalanced input voltage using the SVM algorithm without any compensation and using the SVM algorithm compensated by the SMC based on the reaching law, respectively.
From
Fig. 12
(a), using the conventional SMC algorithm can achieve both output voltage regulation and a higher power factor. However, it also has a drawback since it can only be switch to two modes and cannot change the states according to the running condition. In the dynamic process it requires the system mode switch in a wide range, but for the steady state operation this kind of wide range switch is not conducive to the output performance. As shown in
Fig. 12
(b), the frequency of the current harmonic is mainly located near 1.4kHz. Using the same circuit parameters, but applying the reaching law can solve this problem effectively through changing the switch state to adapt to different working condition. The experimental waveform applying the RLSMC are as shown in
Fig. 13
(a) and
Fig. 13
(b). In addition,
Fig. 13
(a) shows the target voltage undergo a mutation from 80V to 60V, with a response time of 4ms.
Experimental results of using conventional SMC compensation algorithm:(a) AC supply at the grid side and the DC output voltage; (b) spectrum analysis of the Aphase input current.
According to
Fig. 14
(a), using the SVM algorithm without any compensation control approach, to control the MR results in a ripple in the output voltage and a decrease in the power factor at the grid side in the case of unbalanced sinusoidal supply voltages. However, under the same conditions, using the SVM algorithm with compensation by the RLSMC based on the reaching law and NSMC can effectively achieve both output voltage regulation and power factor control, as shown in the
Fig. 13
(b). The experimental results show good agreement with the theoretical analysis.
Experimental results of using SMC compensation based on the reaching law when the threephase input supply voltage is balanced:(a) AC supply at the grid side and the DC output voltage; (b) spectrum analysis of the Aphase input current.
Experimental results when the AC voltage supply is unbalanced: (a) using the SVM algorithm; (b) using the RLSMC algorithm.
However, the RLSMC algorithm proposed in this paper cannot alleviate chattering well, and the simulation results and experiment results are not the same, due to nonideal bidirectional switches operating in the real experiment instead of the simulation environment.
V. CONCLUSION
The novel approach of achieving both a tight DC voltage regulation and direct power factor control by the SVM compensated by the SMC based on the reaching law has been theoretically justified and experimentally verified. A simple singlephase model of a threephase acDC matrix rectifier has been proposed. Its effectiveness has been verified through simulation and experimental results.
The experimental results also demonstrate that applying a sliding mode control based on the reaching law can improve the robustness and dynamic performance of the matrix rectifier. Furthermore, the proposed control method is also capable of wide leading or lagging power factor operation which may broaden its application rang.
BIO
Zhiping Wang was born in Dongguan, Guangdong Province, China, in 1978. He received his B.S. and M.S. degrees from the South China University of Technology (SCUT), Guangzhou, China. He is presently working towards his Ph.D. degree in Electrical Engineering in the School of Electrical Power, South China University of Technology. He is presently working as an Associate Professor at the Guangdong Institute of Automation, Guangzhou, China. His current research interests include highpower density rectifiers and control strategies.
Yunshou Mao was born in Guangzhou, Guangdong Province, China, in 1989. He received his B.S. degree in Electronics and Engineering from the North University of China (NUC), Taiyuan, China, in 2013. He is presently working towards his M.S. degree in Electrical Engineering at the Guangdong University of Technology (GDUT), Guangzhou, China. His current research interests include highpower density rectifiers, multilevel converters and control strategies.
Zhanhu Hu was born in Zhanjiang, Guangdong Province, China, in 1965. He received his B.S., M.S., and Ph.D. degrees in Computer Information Systems Engineering and in Computer Application and Aerospace Control Engineering from the Northwestern Polytechnical University, Xi'an, China, in 1988, 1991, and 1999, respectively. He was a postdoctoral researcher from 1999 to 2001 in Nanjing. He is presently working as a Professor at the Guangdong Institute of Automation, Guangzhou, China. His current research interests include wavelet theory, control engineering and signal processing.
Yunxiang Xie was born in Shaoyang, Hunan Province, China, in 1965. He received his B.S., M.S., and Ph.D. degrees in Electrical Engineering from Xian Jiaotong University, Xian, China, in 1985, 1988, and 1991, respectively. He was a postdoctoral researcher from 1991 to 1994. He was an Associate Professor from 1994 to 1999, and has been a Professor since 1999, in the School of Electrical Power, South China University of Technology, Guangzhou, China. His current research interests include power electronics and control, including highpower density rectifiers, multilevel converters, active filters, and distributed power quality and control strategies.
Kolar J. W.
,
Friedli T.
,
Rodriguez J.
,
Wheeler P. W.
2011
“Review of threephase PWM AC–AC converter topologies,”
IEEE Trans. Ind. Electron.
58
(11)
4988 
5006
DOI : 10.1109/TIE.2011.2159353
Wheeler P. W.
,
Rodriguez J.
,
Clare J. C.
,
Empringham L.
,
Weinstein A.
2002
“Matrix converter: a technology review,”
IEEE Trans. Ind. Electron.
49
(2)
276 
288
DOI : 10.1109/41.993260
Holmes D. G.
,
Lipo T. A.
1992
“Implementation of a controlled rectifier using ACAC matrix converter theory,”
IEEE Trans. Power Electron.
7
(1)
240 
250
DOI : 10.1109/63.124596
Xu Z.
,
Yin G. X.
,
Xu D. G.
2011
“A new bidirectional AC–DC converter using a matrix converter topology,”
Transactions of China Electrotechnical Society
26
(8)
64 
70
Venturini M.
,
Alesina A.
“The generalized transformer: a new bidirectional sinusoidal waveform frequency converter with continuously adjustable input power factor,”
in IEEE Power Electronics Specialists Conference (PESC)
1980
242 
252
Roy G.
,
April G. E.
“Cycloconverter operation under a new scalar control algorithm,”
in 20th Annual IEEE Power Electronics Specialists Conference (PESC)
Jun. 1989
Vol. 1
368 
375
Casadei D.
,
Serra G.
,
Tani A.
,
Zarri L.
2002
“Matrix converter modulation strategies: a new general approach based on spacevector representation of the switch state,”
IEEE Trans. Ind. Electron.
49
(2)
370 
381
DOI : 10.1109/41.993270
Liu X. Y.
,
Zhou B.
,
Xian H. Q.
,
Shi M. M.
2012
“Investigation on grid side power factor of two stage matrix converter,”
Transactions of China Electrotechnical Society
27
(12)
71 
78
Cai W.
,
Yang X. J.
,
Gong Y. M.
2006
“Research on power characteristics of matrix rectifier,”
Automation of Electric Power Systems
30
(8)
27 
31
Wang Z. P.
,
Xie Y. X.
,
Yang K.
,
Luo M.
2014
“Feedforward fuzzy compensation control of matrix rectifier in abnormal conditions,”
Electric Power Automation Equipment
34
(5)
98 
104
Lu X. N.
,
Sun K.
,
Li G.
2010
“Control method analysis of grid side power factor in two stage matrix converter,”
Transactions of China Electrotechnical Society
25
(10)
108 
114
Zhu J. D.
,
Xu Z.
,
Jiang B. H.
,
Zhang C. H.
“Closeloop control of an ACDC matrix converter for automotives,”
in IEEE Electrical Power and Energy Conference (EPEC)
Oct. 2011
426 
431
Deng W. L.
,
Yang X. R.
,
Zhu J. L.
2006
“Study of closed loop control based on double synchronous rotating frame for twostage matrix converter under unbalanced load,”
Proceedings of the CSEE
26
(10)
70 
75
Nguyen H. M.
,
Lee H.H.
,
Chun T.W.
2011
“Input power factor compensation algorithms using a new directSVM method for matrix converter,”
IEEE Trans. Ind. Electron.
58
(1)
232 
243
DOI : 10.1109/TIE.2010.2044736
Liu X.
,
Zhang Q. F.
,
Hou D. L.
2013
“Sliding mode variable structure control of matrix rectifiers,”
Transactions of. China Electrotechnical Society
28
(4)
149 
164
Venturini M.
,
Alesina A.
“The generalized transformer: a new bidirectional sinusoidal waveform frequency converter with continuously adjustable input power factor,”
in IEEE Power Electronics Specialists Conference (PESC)
1980
242 
252
Alesina A.
,
Venturini M.
“Intrinsic amplitude limits and optimum design of 9switches direct PWM AC–AC converters,”
in IEEE Power Electronics Specialists Conference(PESC)
Apr. 1988
Vol. 2
1284 
1291
Kiran M. K.
,
Srinivasulu A.
2012
“Matrix converter using svm method with input power factor compensation,”
IOSR Journal of Electrical and Electronics Engineering
2
(4)
1 
10
DOI : 10.9790/16760240110
Liu X.
,
Zhang Q. F.
,
Hou D. L.
,
Wang S. Y.
2013
“Improved space vector modulation strategy for ACDC matrix converter,”
Journal of Power Electronics
13
(4)
647 
655
DOI : 10.6113/JPE.2013.13.4.647
Tan S. C.
,
Lai Y. M.
,
Tse C. K.
2008
“Indirect sliding mode control of power converters via double integral sliding surface,”
IEEE Trans. Power Electron.
23
(2)
600 
611
DOI : 10.1109/TPEL.2007.915624