Ijmctr042006

International Journal of Modern Communication Technologies & Research (IJMCTR)
ISSN: 2321-0850, Volume-4, Issue-10, October 2016
14 www.erpublication.org

Abstract— This paper presents a review of the saliency-based
soft sensors. In particular estimator based on rotating or
alternating carrier signal injection and methods based on
inductance variations, like INFORM, are presented. The soft
sensors structure and their design are discussed.
Index Terms— Low speed region, Permanent magnet
synchronous motor, Sensorless control, Signal injection.
I. INTRODUCTION
Permanent magnet synchronous machine (PMSM) drive
have been increasingly applied in a wide variety of industrial
applications by replacing classic dc drives. The reason comes
from the special advantages of PMSM such as inherent high
power density, high efficiency, simple structure and absence
of filed losses. To achieve high performance field-oriented
control, accurate rotor position information, which is usually
measured by rotary encoders is necessary. However, the cost
of a sensor may exceed the cost of a small motor in some
applications. Also, the presence of the mechanical sensors not
only increases the cost and complexity of the total material
with additional wiring but also reduces its reliability with
additional sensitivity to external disturbances. In addition, it
may be difficult to install and maintain a position sensor due
to the limited space and rigid work environment with high
vibration or high temperature. Therefore, the idea is to replace
the mechanical sensor by a soft sensor which offers a number
of attractive properties one of them being a low cost
alternative to hardware speed measurement used in classical
motor drives [1–3].
During the years, researchers have developed different
sensorless techniques. These methods were classified under
two main categories:
 Methods suitable for standstill and low speed
region,
 Methods adequate for high speed region.
At high speed, the position can be estimated from the first
harmonic of the back-emf or from the mathematical model of
the drive. Several techniques inspired from control theory
[4–6], such as adaptive observers [7–10], reference models
[11–13], and extended Kalman filter [14].
At standstill, the PMSM is unobservable since the
back-emf are zero. Therefore, classical methods based on the
fundamental wave of the back-emf fail in this region.
At low speed, the rotor position can be estimated by using
inductance variations due to magnetic saturation and/or
Ines Omrane, University of Poitiers, Laboratoire d'Informatique et
d'Automatique pour les Systèmes, Poitiers, France.
geometrical effects of PMSM. This is achieved by injecting
a high frequency signal in the stator windings of the main
generator. This signal can be a voltage or a current signal,
having a frequency other than the fundamental.
Correspondingly, the current or the voltage response,
containing information on the anisotropy, can be used for
detecting the rotor position.
The injection of HF currents presents some drawbacks such
as the bandwidth estimation is limited by the current
controller bandwidth. This problem disappears when a
voltage is injected. In this case, the estimation bandwidth is
determined by the observer bandwidth.
Several techniques have been developed in order to offer a
solution to the estimation problems at low speeds. Thus, the
startup of the motor can be provided in a closed loop. In this
paper, the principle of the most popular sensorless techniques
is presented.
II. INFORM METHOD
The INFORM method " INdirect Flux detection by Online
Reactance Measurement" was first introduced by Shroedl
[15]. This method was based on real-time inductance
measurements using saliency and saturation effects. A
sequence of discrete voltage pulses is injected into the
machine, and the rotor position estimate is deduced from the
difference in the stator inductance.
The stator voltage and stator flux linkage equations in
space phasors can be described in the following way:
s
s s s
s s s pm
d
u R i
dt
L i

 
 
 
(1)
For a sufficiently high frequency of injection, it can be
considered that the stator resistance impedance is negligible
compared to the inductances impedances. Furthermore a
measurement at standstill induced no back-emf. With this
assumption the complex INFORM inductance can be defined
[15]:
s
INFORM
s
u
L
di
dt
 (2)
where LINFORM is the complex INFORM reactance, is is the
complex stator current and t is the time.
The current change with respect to the voltage depends on
A review of saliency-based soft sensors for
permanent magnet synchronous machines
Ines Omrane

A review of saliency-based soft sensors for permanent magnet synchronous machines
the rotor position. In [16], the authors propose to calculate the
incremental inductance in real time and then to deduce the
rotor position. This inductance can be given by equation (3)
by neglecting the emf at standstill or low speed.
 , s s s
s
u R i
L i
di
dt


 (3)
We note that the inductance can be calculated from the
voltage and the measured current. Furthermore, it can be
written as follows
           0 1 2, cos cos 2L i L i L i L i     (4)
Equation (4) represents the first three terms of the Fourier
series. By identifying the equations (3) and (4) and using a
correspondence relationship between the inductance, the
current and position, we can estimate the rotor position.
In the INFORM method, the current variations caused
by the injection voltages are directly used for estimating
the rotor position. The machine parameters will not be
involved in the calculation. Therefore, the INFORM method
is one kind of parameter independent sensorless control
method. However, the main drawback with INFORM is the
unsatisfactory accuracy since current variations are the only
variables used for the estimation. Therefore, the performance
of the position estimation can be dramatically affected by the
accuracy of the injected voltage vectors on the machine
terminals.
III. CURRENT IMPULSES METHOD
This method It is based on the injection of voltage pulses in
the motor phases and the measurement of the related current
peeks. The impulses imposed must be sufficiently large to
produce a change in the saturation state of the engine. When a
voltage step is applied to the phase 'a', the current response
can be written as:
 
1
s
a
R t
La
a
s
u
i e
R

 
  
 
 
(5)
By applying positive and negative voltage pulses on phase
'a', the related currents i+
a and i-
a can be measured respectively.
The rotor position is then determined from the difference
between these two peaks. For positive impulse, the stator flux
is in the same direction as with the mutual flux. Therefore, the
total flux a can be given by
a am aa    (6)
While for a negative impulse of same duration, the stator
flux is in the opposite direction of the mutual flux and
consequently the resulting total flux can be written as
a am aa    (7)
Therefore, the saturation is lower compared to the case of a
positive impulse. The difference of amplitude between the
two current peaks allows to calculate the half of the electrical
period, thus the rotor position. The same technique is applied
to the phase 'b'.
The current differences ai and bi can be written as
follows
a a a
b b b
i i i
i i i
 
 
  
  
(8)
Since ai and bi vary in a sinusoidal way for the different
rotor positions, then the estimated position can be obtained by
simple trigonometric calculations.
This technique has been used in several publications
[17–18], and it was considered as a simple method allowing
the discriminate between a north and a south pole of the
machine. However, the current impulses injection generates a
large amount of noise which must be reduced in order to
obtain a good accuracy.
IV. HIGH FREQUENCY SIGNAL INJECTION
The high frequency method for estimating the rotor
position was a hot research topic due to improvements in
power technologies. This method exploits the different
anisotropies of the machine, such as magnetic saliency,
saturation and eddy current to detect the rotor position. This
method has the same principle as a mechanical sensor. It is
based on the injection of a high frequency voltage (or
current). The measured current (or voltage) contains
information about the rotor position. Different types of high
frequency signals can be injected. the most used ones are the
rotating high frequency signal and the pulsating high
frequency signal.
The magnitude of the injected signal should be selected
such that the rotor remains at standstill during the estimation
phase. The frequency of this signal may vary between 100 Hz
and 4KHz. The most used frequency is 500Hz.
A. Rotating high frequency carrier injection
In 1995, the rotating voltage injection has been tested on
the interior PMSM [19]. Many saliencies have been exploited
to extract the rotor position. Generally the injected signal is a
HF balanced three phase voltage (or current) which can be
carried by the (α, β) axes [20–22] or the (d, q) axes [23-25]. It
has been proved that the current injection requires that the
bandwidth of the current regulators must be greater than the
injection frequency. Therefore, the injected signal is often a
high frequency voltage. The PMSM can be modeled in the dq
reference frame by the following set of equations
0
0
e
d d ds
q q qs
e
d
u iR dt
u iR d
dt




 
       
        
       
 
 
(9)
where

0
0 0
d d d pm
q q q
L i
L i
 

      
       
     
(10)
In the fixed reference frame, the PMSM model can be given
by
s
u i d
R
u i dt
  
  


     
      
     
(11)
where
cos
sin
e
pm
e
i
L
i
 

 
 

 
     
        
    
(12)
Let
2
q d
diff
L L
L

 (13)
and
2
q d
moy
L L
L

 (14)
So we have
cos2 sin2
sin2 cos2
moy diff e diff e
diff e moy diff e
L L L
L
L L L

 
 
  
         
(15)
The injected signal can be given by the following equation
where ui is the magnitude of the injected voltage which should
be properly selected, such that the rotor remains stationary
upon the injection.
 
 
cos
sin
ii
i
i i
tu
u
u t




  
   
    
(16)
Then the stator currents can be given by
   
   
sin sin 2
cos cos 2
pi i ni e ii
i
i pi i ni e i
I t I ti
u
i I t I t


  
  
   
   
      
(17)
where Ipi and Ini are respectively the direct and the inverse
components of the stator currents.
They can be expressed as follows:
 2 2
i moy
pi
i moy diff
u L
I
L L


(18)
 2 2
i diff
ni
i moy diff
u L
I
L L


(19)
We can note that Ini contains information about the rotor
position. In order to extract this information, Ipi must be
completely eliminated. In [22], Wang proposed the following
block diagram given by Fig. 1 where filters and an integral
proportional corrector are used for the extraction of the rotor
position.
Fig. 1. Block diagram of the rotor position estimation
where
e
i
~ is calculated as follows
   
    
ˆ ˆcos 2 sin 2
ˆsin 2 sin 2
e
ni e i ni e i
ni e e ni e
i i t i t
I I
 
   
  
    
  
(20)
Then, the estimated rotor position can be given by
 ˆ
e
e p iK K dt i
    (21)
B. Pulsating high frequency carrier injection
The pulsating high frequency injection scheme makes also
use of the saliencies present in the machine to extract the
position. Usually, a HF signal is injected in one axis of the
reference frame. If the injected signal is a voltage, then the
carrier current response along the axis orthogonal to the
injection axis contains the rotor position information.
Therefore, a high frequency voltage is added to the d-axis
control output as follows:
 cos
0
di i
i
qi
u t
u
u
   
   
   
(22)
By injecting a carrier high-frequency signal, the machine can
be considered as a RL load given by the following equations:
 di
di s di d s i d di d di
di
u R i L R j L i z i
dt
     (23)
 qi
qi s qi q s i q qi q qi
di
u R i L R j L i z i
dt
     (24)
where diu and qiu are the stator voltages at HF, dii and qii are
the stator currents at HF, zd and zq are the d- and q-axes
impedances, respectively, ωi is the frequency of the injected
signal.
Equations (23) and (24) may be given by:
0
0
di d di
qi q qi
u z i
u z i
     
     
     
(25)

A review of saliency-based soft sensors for permanent magnet synchronous machines
Since
ˆcos sin
ˆsin cos
di die e
qi qie e
i i
i i
 
 
   
    
        
(26)
Then
1ˆ ˆ ˆcos2 sin2
2 2 2
d q d q d q
di e di e qi
d q
z z z z z z
i u u
z z
 
     
    
  
(27)
and
1ˆ ˆ ˆsin2 cos2
2 2 2
d q d q d q
qi e di e qi
d q
z z z z z z
i u u
z z
 
     
     
  
(28)
Let
diff d qz z z  (29)
and
2
q d
moy
z z
z

 (30)
Then
1 1 1ˆ ˆ ˆcos2 sin2
2 2
di moy diff e di diff e qi
d q
i z z u z u
z z
 
  
    
  
(31)
and
1 1 1ˆ ˆ ˆsin2 cos2
2 2
qi diff e di moy diff e qi
d q
i z u z z u
z z
 
  
     
  
(32)
The carrier current response along the q-axis is then given
by:
 sin2ˆ sin
2
i e
qi diff i
i d q
u
i z t
L L




 (33)
For a sufficiently high frequency of injection, it can be
considered that the stator resistance impedance is negligible
compared to the inductances impedances
 
d s i d i d
q s i q i q
diff i diff i d q
z R j L j L
z R j L j L
z j L j L L
 
 
 
  
  
  
(34)
Then, qiiˆ can be written by:
 sin2ˆ sin
2
i e
qi diff i
i d q
u
i L t
L L




 (35)
qiiˆ depends on the position error e
~
, so a synchronous
demodulation is required to calculate the proportional signal
to the estimation error of the rotor position. The use of analog
filters allows the isolation of the term that contains
information about the position and the elimination of
undesirable terms. In general, the diagram can be given by
Fig.2. The band pass filter (BPF) is centered on the carrier
frequency. The resulting current is then multiplied by
 tisin . At the end, a low pass filter is used to get back the
signal
e
i
~ approached by equation (36)
 ˆ sin sin2
4e
i diff
qi i e
i d q
u L
i LPF i t
L L
 


  (36)
Fig. 2. Demodulation scheme used to obtain the position
error signal
Suppose that the rotor position estimation error is so small
( ee 
~
2
~
2sin  ). Then,
2e
i diff
e
i d q
u L
i
L L



 (37)
Finally, the rotor speed can be estimated using a PI controller
ˆ
e e
e p ii i dt 
     (38)
where p and i are proportional and integral gain of the
controller. The rotor position estimate is given by:
ˆ ˆe edt   (39)
V. SIMULATION RESULTS
In this section we present the simulation results of a soft
sensor based on signal injection technique. We inject a HF
alternative voltage. Indeed, this method was more accurate.
Fig. 3 shows the simulation results where (a) actual and
estimated mechanical speed, (b) actual and estimated electric
position, (c) speed estimation error, (d) position estimation
error. This soft sensor provides a good estimate of the speed
and position of the MSAP at low speed and even at a
standstill.

VI. CONCLUSION
In this paper, most of the used sensorless techniques have
been described for standstill and low speed. The position
estimation method based on the injection of a rotating or a
pulsating high frequency signal is presented. The basics of the
scheme, the used filtering and the observer structure are
shown. The INFORM method based on inductance variations,
is also presented.
REFERENCES
[1] R. Abdelli, D. Rekioua, T. Rekioua (2011), “Performances
improvements and torque ripple minimization for VSI fed induction
machine with direct control torque”. ISA Trans, Vol. 50, Iss. 2, pp.
213-219.
[2] A.Y. Achour, B.Mendil, S. Bacha, I. Munteanu (2009),
“Passivity-based current controller design for a permanent-magnet
synchronous motor”. ISA Trans, Vol. 48, Iss. 3, pp. 336-346.
[3] B. Zhang, Y. Pi, Y. Luo (2012), “Fractional order sliding-mode control
based on parameters auto-tuning for velocity control of permanent
magnet synchronous motor”. ISA Trans, Vol. 51, Iss. 5, pp. 649-656.
[4] Ph. Bogaerts, A. VandeWouwer (2003). “Software sensors for
bioprocesses”. ISA Trans, Vol. 42, Iss. 4, p.p. 547-558.
[5] R. F. Escobar, C. M. Astorga-Zaragoza, A. C. Tellez-Anguiano, D.
Juarez-Romero, J. A. Hernandez, G. V. Guerrero-Ramirez (2011).
“Sensor fault detection and isolation via high-gain observers:
application to a double-pipe heat exchanger”. ISA Trans, Vol. 50, Iss.
3, p.p. 480-486.
[6] S. R. Vijaya Raghavan, T. K. Radha krishnan, K. Srinivasan (2011).
“Soft sensor based composition estimation and controller design for an
ideal reactive distillation column”. ISA Trans, Vol. 50, Iss. 1, p.p.
61-70.
[7] H. Kubota, K. Matsuse (1993). “DSP-based speed adaptive flux
observer of induction motor”. IEEE Trans Ind Appl, Vol. 29, Iss. 2,
p.p. 344–348.
[8] H. R. Karimia, A. Babazadehb (2005). “Modeling and output tracking
of transverse flux permanent magnet machines using high gain
observer and RBF neural network”. ISA Trans, Vol. 44, Iss. 4, p.p.
445-456.
[9] M. Hinkkanen, M. Harnefors, J. Luomi (2010). “Reduced-order flux
observers with stator-resistance adaptation for speed-sensorless
induction motor drives”. IEEE Trans Power Electron, Vol. 25, Iss. 5,
p.p. 1173-1183.
[10] S. Zheng, X. Tang, B. Song, S. Lu, B. Ye (2013). “Stable adaptive PI
control for permanent magnet synchronous motor drive based on
improved JITL technique”. ISA Trans, Vol. 52, Iss. 4, p.p. 539-549.
[11] G. Yang, T. Chin (1993). “Adaptive-Speed identification scheme for a
vector-controlled speed sensorless inverter-induction motor drive”.
IEEE Trans Ind Appl, Vol. 29, Iss. 4, p.p. 820-825.
[12] G. Madadi-Kojabadi (2005). “Simulation and experimental studies of
model reference adaptive system for sensorless induction motor drive”.
Simul Model Pract Theory, Vol. 13, Iss. 6, p.p. 451-464.
[13] T. Orlowska-Kowalska, M. Dybkowski (2010). “Stator-current-based
MRAS estimator for a wide range speed-sensorless induction-motor
drive”. IEEE Trans Ind Electron, Vol. 57, Iss. 4, p.p. 1296-1308.
[14] D. Xu, S. Zhang, J. Liu (2013). “Very-low speed control of PMSM
based on EKF estimation with closed loop optimized parameters”. ISA
Trans, Vol. 52, Iss. 6, p.p. 835-843.
[15] M. Schroedl (1996). “Sensorless control of AC machines at low speed
and standstill based on the INFORM method”. Industry Applications
Conference IEEE-IAS Annual Meeting, Vol. 1, p.p. 270-277.
[16] H. Gao, F.R. Salmasi, M. Ehsani (2004). “Inductance model-based
sensorless control of the switched reluctance motor drive at low speed”.
IEEE Trans Power Electron, Vol. 19, Iss. 6, p.p. 1568-1573.
[17] M. Boussak (2005). “Implementation and experimental investigation
of sensorless speed control with initial rotor position estimation for
interior permanent magnet synchronous motor drive”. IEEE Trans
Power Electron, Vol. 20, Iss. 6, pp. 1413-1422.
[18] M. Tursini, R. Petrella, F. Parasiliti (2003). “Initial rotor position
estimation method for PM motors”. IEEE Trans. Ind. Appl, Vol. 39,
Iss. 6, pp. 1630-1640.
[19] P. L. Jansen, M. J. Corley, R. D. Lorenz (1995). “Flux, position, and
velocity estimation in AC machines at zero and low speed via tracking
of high frequency saliencies”. European Conference on Power
Electronics and Applications EPE, Sevilla, Spain, p.p. 154-160.
[20] D. Saltiveri, A. Arias, G. Asher, M. Sumner, P. Wheeler, L.
Empringham, C. Silva (2006). “Sensorless control of surface-mounted
permanent magnet synchronous motors using matrix converters”.
Electrical Power Quality and Utilization, Journal EPQU, Vol. 12,
Iss.1, pp. 59- 67.
[21] O. Mansouri-Toudert, H. Zeroug, F. Auger, A. Chibah (2012).
“Improved rotor position estimation of salient-pole PMSM using high
frequency carrier signal injection”. International Conference on
Electrical Machines ICEM, Marseille, France, p.p. 761-767.
[22] G. Wang, R. Yang, Y. Wang, Y. Yu, D. Xu (2010). “Initial rotor
position estimation for sensorless interior PMSM with signal
injection”. International Power Electronics Conference IPEC, pp.
2748-2752.
[23] M. Linke, R. Kennel, J. Holtz (2002). “Sensorless position control of
permanent magnet synchronous machines without limitation at zero
speed”. The 28th annual conference of the IEEE industrial electronics
society IECON, Sevilla, Spain, p.p. 674-679.
[24] S. Kim, S. K. Sul (2011). “High performance position sensorless
control using rotating voltage signal injection in IPMSM”. The 14th
European Conference on Power Electronics and Applications EPE.
[25] Y. Jeong, R. D. Lorenz, T. M. Jahns, S. Sul (2003). “Initial rotor
position estimation of an interior permanent magnet synchronous
machine using carrier-frequency injection methods”. IEEE
International Electric Machines and Drives Conference IEMDC.

Ijmctr042006

More Related Content

What's hot (20)

Viewers also liked (17)

Similar to Ijmctr042006 (20)

More from Engineering Research Publication (20)

Recently uploaded (20)

Ijmctr042006