MATEC Web of Conferences 31 , 0 7 0 0 1 (2015)
DOI: 10.1051/ m atec conf/ 201 5 3 10 7 0 0 1
C Owned by the authors, published by EDP Sciences, 2015
Sensorless Control Technology for PMSG base on the Dead-time
Compensation voltage
<DQJ/L\RQJ=KDL)HLD
1
1RUWK&KLQD8QLYHUVLW\RI7HFKQRORJ\&KLQD
2
1RUWK&KLQD8QLYHUVLW\RI7HFKQRORJ\&KLQD
Abstract: In order to improve the speed sensorless-control system of PMSG in low speed performance, this paper
introduces a novel Dead-time compensation control method .Mathematical model is established according to the
Dead-zone of the influence of the voltage source type inverter output voltage. At the same time, the given value of
current regulator output voltage has been fixed based on the established model. Then the stator voltage after
compensationed is applied to the flux estimation, which improves the performance of flux estimation. Finally, the
position and speed of the rotor is estimated based on Back-Electromotive Force, which has Simple algorithm and
good robustness. In order to verify the correctness of theoretical analysis, the experiment was done according to the
new control method. The results proved the correctness and feasibility of this control method.
Key Words: Permanent Magnet Synchronous wind power Generator; Back Electromotive Force; Dead-time
voltage Compensation; Flux Estimate
that the reliability of the system is improved, the cost is
,QWURGXFWLRQ
reduced greatly [5].
Wind energy is widely seen as a substitute for the
In
the
speed
sensorless permanent
traditional non renewable energy due to its large-scale
magnet synchronous wind generator control system,
development
pollution.
flux estimation requires stator voltage. If a given
power
voltage in the SVPWM algorithm is used as electronic
generator has a wide application prospect, because of
voltage. It will cause error about the flux estimation
its compact
and
and reduce the flux estimation accuracy. Finally it have
excellent power system fault adaptive [1-3].Sensorless
a serious impact on the performance of low-speed
control technology is to estimate the rotor position and
permanent magnet motor, because the dead time effect
speed by a specific algorithm, related to the use of
[6].
Permanent
and utilization and no
magnet
structure,
synchronous
reliable
wind
performance
The Back-EMF method is widely used, because of
easily measured variables which is in the motor
winding, such as the stator voltage, stator current,
its
physical concept is
winding flux. The purpose is to realize the high
calculation [7]. The combination of Back-EMF and
precision control of permanent magnet motor [4].A
dead voltage compensation is achieved the precise
series of problems existing in mechanical sensor can be
control
solved by using the speed sensorless technology. So
synchronous generator. In this paper, a method is
of
the
clear,
sensorless
small
amount
permanent
of
magnet
introduced about the sensorless permanent magnet
a
Corresponding author:zhaifei523@163.com
This is an Open Access article distributed under the terms of the Creative Commons Attribution License 4.0, which permits XQUHVWULFWHGXVH distribution, and reproduction in any medium, provided the original work is properly cited.
Article available at http://www.matec-conferences.org or http://dx.doi.org/10.1051/matecconf/20153107001
MATEC Web of Conferences
d s
e s u s Rs i s
dt
synchronous wind generator control system .First of
all, the
inverter
output
voltage distortion is
(2)
compensated, by the detection of three-phase voltage
The equation(2)on both sides of integral
and three-phase current in the system. Then, according
to
the
motor model,
position deviation
we
calculated
of motor
and
the rotor
s es dt u s Rs is dt
used Back-EMF
Type(3)in the form of a component
method to track estimation of rotor position. Finally,
the
theory is
verified
(3)
by hardware
s es dt u s Rs is dt
(4)
s es dt u s Rs is dt
experiment platform based on TMS320F28335 control
panel.
The calculation process of stator flux using the
0RGHOLQJIRU306*
formula(4)need use not only the stator current signal,
but also the stator voltage signal. So the formula(4)is
$ PDWKHPDWLFDO PRGHO RI%DFN(0) PHWKRG
called
the voltage
model in alpha
beta coordinate
of stator flux linkage. In the process of flux estimation,
306*
the estimation accuracy of the voltage model may be
affected by the pure integrator. This is due to the
Mathematical model is an important means to study
voltage and current signal of DC component and the
the actual physical object. In order to analyze and
initial error and especially when the motor is in low
control of PMSG, the establishment of mathematical
frequency, this kind of influence is more serious.
model with simple and feasible is very necessary.
There
are
generator vector
two kinds
control
of coordinates
system
of
In order to solve this problem, a low pass filter is
in the
introduced to replace the pure integral of basic voltage
permanent
model.
magnet synchronous wind. One is two-phase rotating
The transfer function of the lowpass filter is
coordinate system (d-q coordinate) and the other is
GLP s two-phase stationary coordinate system (alpha-beta
coordinate system).In order to calculate the rotor
1
s c
.where c is the cutoff frequency of the
lowpass filter.
position, We have to get the value of the stator flux
According to equation(4)can be obtained for the
linkage. A simple method of estimating the stator flux
stator flux estimator based on the low-pass filter.
is the voltage model of stator flux in the alpha beta
1
1
s s s es s s u s s Rs is s c
c
1
1
s s es s u s s Rs is s s c
s c
coordinate system and it is also the basis of other flux
estimation method. This method only requires the use
of a stator resistance parameter. So it is easy to be
(5)
realized and has better robustness. When the power
supply frequency is high, this estimation method
q
(B)
has better estimation accuracy especially.
In alpha-beta coordinate
system,
c
N
electromotive force is defined as
es u s Rs is
d
vector induced
i
(A)
S
(1)
The relationship between the stator flux vector and
C
Fig.1 The relationship between the coordinate system PMSG
induced electromotive force vectoris
07001-p.2
ICMEE 2015
According to the Fig.1,the rotor position of PMSG
can be expressed as [8]
s s i arctan
es f c sin es es f c cos (7)
is difficult to estimate.
Assuming the direction of the current does not
phase is expressed as follows, relative to the midpoint
of the busbar O.
v Ao Vd c Vce Vd S A 0.5 (8)
f
0.5signi A Vce Vd 7KHPDWKHPDWLFDOPRGHORIWKHLQYHUWHURXWSXW
In the Back-EMF method, the inverter output
voltage value is used as the system state variables, but
due to the influence of DTCV, the actual output voltage
distortion, eventually lead to serious deviation of the
estimated angle[9].
2
A
VT6
represents the high level of the A phase of the drive
pulse and S A 0 represents the low level of the A phase
of the drive pulse.
Because Vce and Vd is increased with the rise of the
current, effect of time of S A 1 state for Ta, effect of
time of S A 0 state for Ts Ta .
VD1
VD5
V 3
VD
B
VT4
In one switching period, put Vce Vce0 Rce i A ǃ
R
O
C
VD4
VD6
2
In the formula, Vdc is the DC busbar voltage of
VT5
VT3
VT1
n
Vddc
(12)
inverter, Vce is the tube voltage drop of power devices,
Vd is the tube voltage drop of fly-wheel diode. S A 1
YROWDJH
Vddc
(11)
change in a switching period, the actual voltage of A
according to the equation (7)
e
1
VAn VBn VCn 3
direction, the actual phase voltage of A,B,C three-phase
Therefore the motor speed is expressed as
e
(10)
Because of the uncertainty caused by the current
of the alpha beta coordinate axis component.
c i A iB iC 0
V An VBn VCn 0
VnO Where f is rotor flux, es and es is the Back-EMF
2
s
(9)
(6)
Because the Back-EMF es can be divided into type
2
s
V An V AO Vn O
VBn VBO Vn O
V V V
CO
nO
Cn
Vd Vd 0 Rd i A
VD 2
VT2
into the formula
T
v AO Vd c Vce Vd a 0.5 0.5signi A Vce0 Vd 0 T
S
0.5Rce Rd i A
Power device VT1-VT6 can be controlled on and
off by various methods,VD1-VD6 is free wheeling
0
1, i
Where signi A A
, ia .
1, i A 0
diode. Vdc is the DC bus voltage, n is a DC link neutral
(13)
0 represent the
current from the inverter to motor.
Therefore, relative to the neutral point O of the
motor, A,B,C three points voltage and current equations
Ta
Ts
(12),the average voltage can be expressed as
Fig.2 Three phase bridge inverter main circuit diagram
point potential, O motor winding neutral point.
and s A Therefore, VAO has moved with the switch state and
the current direction[10].
respectively
Similarly, B and C phase voltage can be expressed as:
07001-p.3
MATEC Web of Conferences
T
v BO Vd c Vce Vd b 0.5 0.5signiB Vce0 Vd 0 TS
0.5Rce Rd iB
T
vCO Vd c Vce Vd c 0.5 0.5signiC Vce0 Vd 0 T
S
0.5Rce Rd iC
power device turn-off time, Td stands for the dead time,
(14)
Vce-Vd is too small to be ignored.
7KH HIIHFWRI '7&9DQGDQHZYROWDJH
(15)
FRPSHQVDWLRQVWUDWHJ\
Type (9)~(11) into the formula (13)~(15) can be
obtained:
v An we must know the motor phase current direction from
1
Vd c Vce Vd 2Ta Tb Tc
3
TS
equation(20).In this paper, we used the indirect method
for determining the direction of the current, which is
Vce0 Vd 0
2signi A signiB signiC (16)
6
0.5Rce Rd i A
v Bn 1
Vd c Vce Vd 2Tc Tb Ta
3
TS
1
Vd c Vce Vd 2Tc Tb Ta
3
TS
based on the concept of vector. The direction of the
current is according to the angle between current and
voltage vector (power factor) to indirectly determine.
The output of the inverter for three-phase current is
iAǃiBǃiC, Three phase voltage is uAǃuBǃuC. Firstly,
Vce0 Vd 0
2signiB signi A signiC (17)
6
0.5Rce Rd iB
vCn Accurately estimate of the inverter output voltage,
CLARK transform is applied to three phase current and
phase voltage.
Vce0 Vd 0
2signiC signiB signi A (18)
6
0.5Rce Rd iC
The above formula can be simplified as:
1
*
'
V AO V An Vn O V AO V AO 2 Rce Rd i A
1
*
'
VBO VBn Vn O VBO VBO Rce Rd iB
2
V V V V * V ' 1 R R i
CO
Cn
nO
CO
CO
ce
d C
2
1
1 i A
is 2 1 2 2 i
i 3
3 B s 3 0
i 2
2 C (21)
1
1 u A
u s 2 1 2 2 u B u 3
3
s 3 0
u 2
2 C (22)
(19)
Then, two phase currents is and is are transformed
into the coordinate system, which is based on the stator
'
V AO
V ' 2signi A signiB signiC '
'
VBO V 2signiB signiC signi A V ' V ' 2signi signi signi C
B
A
CO
voltage vector for the d axis synchronous rotating
(20)
coordinates.
isd cos" sin " is i sq sin " cos" is *
2TA TB TC
1
V AO ! Vd c
3
Ts
2TB TA TC
1
*
VBO ! Vd c
3
Ts
2TC TB TA
1
*
V
V
!
CO 3 dc
Ts
Where " is the angle between the vector and the A
u s
axis. " arctan
u s
Power factor angle is the angle between the current
and voltage vector, which is obtained by isd and isq.
i
arctan sq DŽ
Where
2 To ff T o nTd
1
Vceo Vd o V ' Vd c
6
T
isd
The relationship between voltage vector and current
,Ton
vector is shown in Fig.3.
stands for power device turn-on time, Toff stands for
07001-p.4
ICMEE 2015
(3) The formula (22) and the calculated voltage error
In the actual system, the stator voltage vector
angle " is known. Power factor angle can be real time
can calculate the real value of A-B axis voltage.
1
1 V $ AO
%u s 2 1 2 2 $ V
%u 3
3 BO s 3 0
V $ 2
2 CO estimation through the instantaneous value of threephase voltage and three-phase current. According to
Fig.3,the angle i between the current vector and
the axis is expressed as: i " (23)
u s& u s %u s
&
u s u s %u s
Bias voltage compensationis determined by the
current direction. However, the three-phase current
Finally, the formula(23)and type(6)-(8)can be
direction is determined by the current vector
angle i .The relationship between i and three-phase
calculated by permanent magnet synchronous wind
current direction as shown in Equation 4[10].
power generator rotor position and speed.
([SHULPHQWDO5HVXOWV
This paper established the hardware platform of the
q
d
speed sensorless vector control system for PMSG
based on TMS320F28335 control board.Fig.5 shows
"
the control block diagram and Fig.6 shows motor
control and test platform.
Fig.3 Relationship between voltage and current
#
#
Back-EMF
C
us &
us &
r
PWM
Inverter
us
11#
6
r
+
%us
+
%us
Dead voltage
compensation
3s
2s
PMSG
U AB
U BC
U CA
id fed
A
Wit
h
coin the C
cid axis
en
ce
SVPWM
2s
r
-
us
2r
iq fed -
PI
0
6
With the B axis
coincidence
7#
6
PI
PI
fed -
With the A axis
coincidence
B
Vdc
Ipark transform
ref
2
5#
6
Non controlled
rectifier
Power grid
ia
ib
ic
ia
ib
2#
3
2r
M
Three-phase current
i
The wind
motor
ic
2s
Park transform Clark transform
Fig.4 Relationship between i and Three-phase
Fig.5 Consider the impact of dead zone sensorless PMSG
vector current direction
According the above analysis, method is given as
follows, which is three phase voltage dead zone
compensation.
(1) A ǃ B ǃ C three-phase current direction is
determined by i and Fig.4. An angle threshold is
introduced to the process and when this angle is less
than the threshold value of angle of three-phase voltage
PMSG Motor
PMSG Driver
compensation, this point of view is the difference
between i and boundary point and Boundary point.
Fig.6 The actual PMSG test platform control system block
This can avoid the error estimate of the error caused by
diagram
the current direction.
In order to verify the validity of this control method,
(2) According to the current direction and formula (20),
the following experiments were studied. PMSG driver
we can calculate the voltage error V AO
$
$ ǃ VBO
$ ǃ VCO
is composed by four functional blocks: Rectifier
caused by the dead time.
module, IPM module, DSP module, and interface
07001-p.5
MATEC Web of Conferences
circuit. The controller is based on DSP TMS320F28335
in
the
inverter
is
PM75RL1A120.The
/ ra d (0.036# /div˅
produced by Texas Instument. The type of IPM module
main
spectifications of PMSG and control parameters in the
experiment are given in Table 1.
The estimation of the
rotor position
The actual rotor
position
t/ms(80ms/div)
Tab.1 Specifications of tested PMSG and control parameters
(a)Before compensation
Value
Rated Speed
1200r/min
Rated Power
0.88Kw
Rated Torque
4.5Nm
Rated voltage
160V
Rated current
5.5A
/ r a d (0.036 # /div˅
Item
The estimation of
the rotor position
The actual rotor
position
t/ms(80ms/div)
Rs
2.875 '
Ls
0.0085mH
P
3
Ts
0.0004s
ton
1us
toff
1us
td
6.4us
Vs
2V
(b)After compensation
Fig.7 Comparing the actual rotor position and the estimated
0.5v/div
rotor position when motor 30r / min
Voltage without compensation usa
&
Voltage with compensation usa
t/ms(40ms/div)
When the motor is at low speed, the impact of dead-
Fig.8 After correction waveforms when motor 50r / min
time effect is very clear. In order to verify the effect of
/ rad (0.036# /div˅
the dead-time compensation, the reference speed is
80r/min, The test result is shown in Fig.6.
Fig.7. shows a comparison of rotor position of
before
and
after
the
dead
voltage
(157mA/div)
waveform
compensation, when the motor is in low speed
(30r/min). We can see that the estimation of the rotor
The estimation of the
rotor position
The actual
ct
rotor
position
t/ms(40ms/div)
position has obvious distortion in Fig.7.a and the
position of the rotor without waveform distortion in
(a)Before compensation
Fig.7.b.Fig.8 is the voltage waveform before and after
compensation, when the motor is in 50r/min.
07001-p.6
ICMEE 2015
(157mA/div)
/ rad (0.036#/div˅
$FNQRZOHGJHPHQWV
This project is sponsored by the National Natural
Science Fund(Project Number:51207003).I greatly
The estimation
n of the
rotor position
The actual rrotor
position
acknowledge this support.
5HIHUHQFHV
t/ms(40ms/div)
1. Chinchilla M, Arnaltes S, Burgos J C, et al(2006).Control
(b)After compensation
of permanent-magnet generators applied to variable speed
Fig.9 Rotor position and current waveforms
wind-energy systems connected to the grid[J].IEEE
when the load motor 50r / min
Transactions on Energy Conversion,21(1):130-135.
Fig.9.shows a comparison of rotor position of
2. Zhao Rende,Wang Yongjun,Zhang
waveform and current waveform before and after the
Jiasheng(2009).Maximum power point tracking control of
Dead-time voltage compensation, when the load motor
the wind energy generation system with direct-driven
is in low speed(50rpm).The rotor position estimation
permanent synchronous generators[J].Proceedings of the
value had great phase deviation before dead time
CSEE, 29(27):106-111.
compensation, but after the compensation and under
3.
J.S.Thongam,M.Tarbouchi,e
tal(2012).An
Optimium
light load condition, the motor still run well all the time.
Speed MPPT Controller for Variable Speed PMSG Wind
System recovery time is about 200ms,the estimation of
Energy Conversion Systems,38th Annual Conference on
the rotor position is accurate and the current waveform
IEEE Industrial Electronic Society,4293-4297.
is good.
4.
Consoli,
Giuseppe
Scarcella,
Antonio
Testa(2001).Industry application of zero-speed sensorless
&RQFOXVLRQV
control techniques for PM asynchronous motors. IEEE
The output voltage of the inverter model is
Trans. Ind.Appl. 37(2):13-521.
established according to the permanent magnet
synchronous
Alfio
generator
system.
Based
on
5. Sergeant P,Dupre L, et al(2010).Losses in Sensorless
the
Controlled
Permanent
Magnet
Synchronous
compensation of voltage, sampling current and Back-
Machines[J].IEEE Transaction on Magnetics , 46(2):590-
EMF
593.
algorithm,
a
novel
Dead-time
voltage
compensation method is proposed. The experimental
6. W.Qiao,W.Zhou,J.M.Aller,and R.G.Harley(2008).Wind
results is reasonable. Although the Back-EMF method
speed estimation based sensorless output maximization
can estimate the rotor position of the motor, but the
control for a wind turbine driving a DFIG, IEEE Trans.
estimation accuracy has a strong dependence on the
Power Electronics, vol. 23, no.3,1156-1169.
parameters of the motor and the motor speed. When the
7.
Tong
Li,Zou
Xudong,Feng
system is subjected to the dead voltage, the rotor
sensorless
position estimation is very easy to deviate from the
synchronous
actual position and make the system unstable. The
algorithm[J].Transactions
control algorithm can effectively improve the stability
Society,28(3):17-26.
control
scheme
generators
Shushuai,etal(2013).A
for
using
of
permanent
predictive
China
magnet
dead-beat
Electrotechnical
of the PMSG speed sensorless control system and the
8. Li Zhengxi,Yang Liyong(2013).AC and DC speed control
estimation accuracy of rotor position. Especially, in the
system[M].Beijing: Publishing House of Electronics
low speed region for PMSG, the control precision is
Industry,2:261-268.
better. But when the load of the motor increases, the
9. Yang Li-yong,Chen Zhigang,Chen Weiqi.(2012).Out
control method needs to be improved.
Voltage Model of VSI-Inverter and a Novel DeadTime
07001-p.7
MATEC Web of Conferences
CompensationMethod[J].Transactions
of
China
Electrotechnical Society,27(1):183-185.
10. Kan Lin and Z.Q.Zhu(2014).Online Estimation of the
Rotor
Flux
Linkage
and
Votage-Source
Inverter
Nonlinearity in Permanent Magnet Synchronous Machine
Drives[J]. IEEE Tansactions on Industrial Electronica,
29(1):419-421.
07001-p.8