2019 IEEE Conference on Power Electronics and Renewable Energy (CPERE)

Robust Predictive Current Control for a Sensorless

IM Drive Based on Torque Angle Regulation
Mahmoud A.Mossa Silverio Bolognani
Electrical Engineering Department Department of Industrial Engineering
Faculty of Engineering,Minia University University of Padova,Via Gradenigo
Minia, Egypt, 61111 6/a, Padova 35131 (Italy)
[email protected] [email protected]

Abstract—The paper presents a robust predictive current The MPC control approaches which are either current
control (PCC) approach for a sensorless induction motor (IM) control based [6,7] or torque and flux control based [8,9]
drive. The proposed PCC is based on the regulation of the have contributed in enhancing the IM drive performance
motor torque angle. The regulation methodology is based on through eliminating the ripples in the controlled variables of
analyzing the relationship between the angular slip frequency torque, flux and current. However, the majority of the
and the torque angle. To reduce the inherent coupling between proposed MPC techniques have ignored the coupling issue
the electromagnetic and mechanical variables of the IM, a between the electromagnetic and mechanical variables of the
modified mathematical model of IM is presented in which the IM (mainly between the rotor flux and the mechanical
nonlinearity degree is effectively reduced and thus perfect
speed), which consequently affects negatively the dynamic
decoupling can be achieved at steady state and transient
operation as well. A robust speed estimator is proposed based
response of the drive especially at transient operation
on a modified flux estimator. In order to show the effectiveness conditions. In order to solve the coupling issue, an attempt is
of the overall control procedure, the drive performance is made in the current paper through presenting new state
tested for a wide range of speed variation. The obtained results variables of IM through which the electric dynamics of the
validate the feasibility of the proposed PCC and verify its motor can be investigated and analyzed.
robustness against the system's uncertainties. It is well recognized that the main challenge of the
sensorless mechanism is to achieve precise estimation for the
Keywords—current control, torque angle, induction motor,
predictive control, sensorless control, nonlinearity.
flux and speed. Previous studies have demonstrated that the
closed loop adaptive speed estimators; either with linear or
I. INTRODUCTION nonlinear configurations have proven their superiority over
the open loop ones [10-14]. However, the speed estimation at
Various sophisticated control approaches have been very low operating frequencies is considered a big challenge.
introduced to obtain better dynamic performances from For this reason, a modified flux estimator current-voltage
induction motor (IM) drives [1-5]. One of these studies has model is presented here. The estimated flux is then used as
concerned with obtaining an IM performance similar to that an input to the model reference adaptive regulator (MRAR)
of the dc motor; this control procedure was entitled as field which finally gives the estimated speed required by the
oriented control (FOC) [1-3]. Another control approach is controller for prediction and speed regulation as well. The
based on preserving a fixed value for the excitation flux effectiveness of the proposed flux estimator is investigated
inside the machine while regulating the torque and speed, through comparing the bode plots of the transfer functions
and to perform this, two hysteresis controllers for torque and for the proposed estimator and voltage model based
flux have been utilized; this approach is known as direct estimator. Through this comparison, it can be realized that
torque control (DTC) [4,5]. Both of these two control the robustness of the proposed flux estimator and
approaches represent the base principle upon which recent consequently the speed estimator is better than the voltage
approaches have relied on. model based estimator.
One of the most recent control strategies for IM drives is The main principle based on which the proposed PCC
the model predictive control (MPC). This control procedure approach is derived entirely lies on regulating the position of
has effectively succeeded in overcoming many drawbacks of the stator current vector respecting to the rotor flux vector; in
the previous two techniques. This can be obviously shown other words, regulating the torque angle. The torque angle
through the elimination of the proportional integrator (PI) has a direct relationship with the angular slip frequency, and
controllers and the hysteresis controllers, in addition to the based on this fact the transfer function which outlines the
elimination of the pulse width modulation (PWM) in the operation of the torque angle PI controller is derived in a
finite state based voltage selection type [6-9]. The MPC systematic manner.
replaces all internal controllers found in FOC and all
hysteresis controllers found in DTC with a simple The paper starts with describing the variables which
mathematical formulation called the cost function which describe the motor dynamics. Then, the proposed PCC
contributes in reducing the system complexity and enables approach is presented and analyzed. After that, the proposed
the integration of extra control objectives to enhance the flux and speed estimators are introduced. Finally, the test
drive performance. results, discussion and conclusion are presented.

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DE MINAS GERAIS. Downloaded on January 16,2024 at 20:36:27 UTC from IEEE Xplore. Restrictions apply.
978-1-7281-0910-7/19/$31.00 ©2019 IEEE 302
 2019 IEEE Conference on Power Electronics and Renewable Energy (CPERE)

II. IM MODEL
, ,
= 2ψ , + 2ψ , = −2 Γ , +
The equivalent circuit of the IM can be represented in
the stationary reference frame as shown in Fig.1. +2 Γ , (7)

i ss s
i rs
Rs Lσs L'σr R'r = 1.5 ∗ ψ ,
,
+i ,
,
+
, ,
i ms
ψ , +i ,
, ,
dψss LM dψsrs Γ Γ + Γ +
u ss jωmeψsrs , , , ,
dt dt =− + ,

+ ψ , u , +ψ , u ,

Fig. 1. IM equivalent circuit

(8)
where μ = , and ∂ = (L L − L ).
2
N   N 
where R 'r   s  R r ; L M   s L m ;
 L σs  L s  L M ;
 Nr   Nr  As can be seen from (5) to (8), the dynamic performance
N 
of the IM can be described by only four differential
L'σr  L r  L m  r s
 , and i rs
  i rs N r N s  ; ψ rs
s
 ψ rs N s N r  . equations instead of five if the classic model is adopted. The
 Ns  four differentials represent the change in the angular speed
The Ns and Nr are the stator and rotor number of turns, (Γ ), the change in the developed torque (Γ ), the change
respectively. Where Ls, Lr and Lm are the stator, rotor and in the rotor flux squared modulus (Γ ), and the change in
magnetizing inductances of the IM, respectively. The the airgap exchange energy (Γ ) which also can be
variable ω me refers to the angular mechanical speed, and i rss considered as indication to the absorbed reactive power. A
, ψ rss the rotor current and flux referred to the stator distinguished merit for the specified state variables (1) to (4)
is that these variables do not require the co-ordinate
side. The superscript s refers to the stator reference transformation which simplifies the control scheme.
frame.
It can be noticed also that (5) and (7) are referring to a
From Fig. 1, the state variables which are used to fully linear relationships between the rotor flux squared
describe the IM dynamics can be expressed in discrete modulus and the exchange airgap energy, and this is true as
forms at instant kTs by the airgap flux is considered as indication to the level of the
internal airgap energy of the IM. On the other hand, the
Γ , =ω , (1) relationships (6) and (7) are not completely linear as the
terms z , and z , initiate a coupling between the
Γ , = 1.5P ∗ ψ , i , −ψ , i , (2) mechanical variables (Γ ,Γ ) and electromagnetic variables
(Γ , Γ ). However, this partial nonlinearity is much less
Γ , = ψ , +ψ , (3) impressive than the nonlinearity in the classic fifth order
model. Moreover, the nonlinearity terms z , and z , can
Γ , = 1.5 ∗ ψ , i , +ψ , i , (4)
be easily treated when applying the predictive controller
through adopting hysteresis based predictive controller
The variables (1) and (2) are used to describe the technique combined with the finite control set strategy for
mechanical properties of the IM, where variables of (3) and selecting the voltages. While, in case of applying the FOC;
(4) are utilized to express the electromagnetic behavior of dealing with the system nonlinearity requires the usage of
the IM, this can be more clarified through taking the multiple PI controllers in addition to the pulse width
derivatives of the (1) to (4) which will give the following modulation (PWM) to provide the voltage vectors, which
four differential equations: increase the complexity of the controller and increases the
computation burdens as well.
,
= = Γ , − T, (5)
III. PROPOSED PCC APPROACH

The proposed PCC is constructed based on regulating

, ,
= 1.5P ∗ ψ , +i , − the position of the stator current vector respecting to the
, ,
rotor flux vector; in other words, regulating the torque
ψ , −i , angle. The space vector diagram which shows the allocation
= of the stator current with respect to the rotor flux when
− Γ , − applying the rotor field oriented control procedure can be
shown through Fig. 2.
−Γ , Γ , + Γ , + ψ , u , −ψ , u , (6)
,

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DE MINAS GERAIS. Downloaded on January 16,2024 at 20:36:27 UTC from IEEE Xplore. Restrictions apply.
303
 2019 IEEE Conference on Power Electronics and Renewable Energy (CPERE)

δ =θ , 1+ (13)

Now, to get the relationship between the torque angle δ

i sr and the angular slip frequency ω , , (13) has to be
ψrr differentiated with respect to the time, then
ωi

ωψr =ω , ∗ 1+ (14)

To determine the parameters of the PI torque angle

Fig. 2. Allocation of stator current vector respecting to rotor flux vector controller, the transfer function which combines the
derivative (variation or error) of the torque angle ( =
Where the superscript r refers to the rotor reference ∗
δ − δ ) as an input and the angular slip frequency
frame, and ω is the angular frequency of the rotor flux
ω , as an output has to be determined and this can be
vector, and ω is the synchronous frequency of the stator
performed as follows:
current vector. Moreover, the torque angle δ has a direct
relationship with the slip angle Ѳ , and through analyzing Taking the Laplace transform of (14); this results in
this relationship; the design of the PI torque angle controller
will be performed in the following section. The output of the , ( )
PI torque angle controller will provide the reference slip = (15)
( )
angular frequency ω∗ , which will be added to the predicted
estimated speed ω , to get the reference angular The operation of the PI torque angle controller can be
frequency of the stator current ω∗, . After that by represented in the Laplace form by
integration the reference angle of the stator current vector
Ѳ∗, can be obtained and then used to get the references of ω∗ , ( )= k + ∗ (δ∗ ( )−δ ( )) (16)
the stator current α-β components i∗ , and i∗ , to be
used by the controller. Then, by comparing the terms of (15) and (16), this results

From Fig.2 and from (2), the developed torque can be δ ( ) = k + ∗ (δ∗ ( )−δ ( ))
evaluated at instant (k+1)Ts by
(17)
By dividing both sides of (17) on δ∗ ( ), then
Γ , = 1.5p ı̅ , ψ , sin δ (9)
( )
∗ + k + = k + (18)
Also, the developed torque can be given in terms of the ( )
angular slip frequency ω , as follows:
From (18), the transfer function that outlines the operation
, of the PI controller will be as follows:
Γ , = 1.5p ψ , + L ı̅ , ψ , sin ω , t
( ) ∗
(10) ( )= ∗ = (19)
( )
where p is the pole pairs and L = L − L ⁄L is the
transient inductance.
To ensure the stability of the system, the characteristic
equation of the denominator in (19) has to be with real
Then, by comparing (9) and (10), this results in
negative roots, thus
L
ı̅ , ψ , sin δ = ψ, sin θ , LL + (L L + L )k + (L L + L )K = 0.0 (20)
LL
+ ı̅ , ψ , θ , (11) It is well realized that the characteristic equation which
describes the operation of a second order differential
When applying rotor field oriented control (RFOC) and for equation can be expressed by
steady state operation, ψ , = L ı̅ , . Then, by dividing
both sides of (11) by ı̅ , ∗ ψ, and after some +2 ω + ω = 0.0 (21)
mathematical abbreviations, this results in
where is the damping coefficient and ω is the natural
sin δ = sin θ 1+ (12) angular frequency.
,
By comparing (20) and (21), the parameters of the torque
angle PI controller shown in Fig. 3 can be calculated by
From (12) and by taking the trigonometric inverse sin ,
this results in ∗ ∗
k = , and k = (22)

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DE MINAS GERAIS. Downloaded on January 16,2024 at 20:36:27 UTC from IEEE Xplore. Restrictions apply.
304
 2019 IEEE Conference on Power Electronics and Renewable Energy (CPERE)

*
is, i*αs, k 1 where i refers to the voltage index and accent ~ refers to the
k 1
ω̂me, k 1 predicted signals. The formulation (27) does not require a
θ i, k 1 i*βs,k 1
weighting balance coefficient as the two terms of the
ωi, k 1
function are with the same nature (current).
δ*k 1 ω*sl,k 1

Till now, the main task of the proposed PCC can be

achieved, as at each instant happened that the estimated
δ̂ k 1
absolute error e which is calculated by (28) deviates
Fig. 3. Role of PI torque controller during implementation
from zero, the control system starts to search for the best
voltage which limit the error increase based on minimizing
The estimated torque angle δ in terms of the new state (27).
variables as follows:
,
δ = tan (23)
, e = (i∗ , − ı̃ , ) + (i∗ , − ı̃ , ) (28)

From Fig. 3, it can be seen that the references of the

stator current components (i∗ , and i∗ , ) are obtained The remaining part in the controller is the flux and speed
using the estimated angle of the stator current vector θ , estimator which is described in the following section.
with the help of the torque angle PI controller. The original
reference of stator current vector ı̅∗, is obtained through IV. PROPOSED FLUX AND SPEED ESTIMATOR
adopting the RFOC principle using the new state variables
introduced previously in (1) to (4) as follows:
Using the IM voltage model to estimate the flux exhibits
∗ ∗
good performance at high speeds but it deteriorates during
ı̅∗, = , ,
(24) low frequency operating conditions, this can be inferred to
∗
, the mismatch in the stator resistance at this condition. As the
estimated flux is main contributor in estimating the speed in
The reference Γ ∗ , represents the torque reference, case of using the model reference adaptive controller
while the reference Γ ∗ , represents the airgap energy (MRAC) which extracts the speed via comparing two values
of rotor flux; one of them obtained directly from the voltage
reference which can be obtained through a PI controller
model and the other is obtained from the current model
taking the error of the rotor flux squared modulus as an which contain the speed within it. The deterioration in the
input, and finally the reference Γ ∗ , is the reference of the rotor flux estimation and consequently the speed estimation
rotor flux squared modulus which is imposed directly. rises mainly from the voltage model due to the change in the
stator resistance (Rs) at low speed operation, and this can be
In the same manner, the torque angle reference is clearly investigated through analyzing the transfer function
calculated by of the voltage model based rotor flux estimator. This transfer
∗
function can be described by the following formulation
δ∗ = tan ∗
,
(25)
, considering a mismatch in the value of Rs by 20 %.
,
The relationship (25) is similar to that one obtained when =
,
applying the RFOC (ψ , = 0.0, and ψ , = ψ, )
using the d-q stator current components as state variables 1+ + jω , ∗ −j (29)
, ,
which in this case is corresponding to the stator current
position that can be defined by By plotting the bode plot of the transfer function (29), it
∗
,
results in the following diagram.
(δ∗ ) = tan ∗ (26)
,
Here (25) is used as stated before to ensure full
decoupling between the electromagnetic and mechanical
variables which cannot be achieved using (26) when
applying the classic form of the RFOC.

After obtaining the references (i∗ , and i∗ , ), then

they can be used by the cost function of the controller which
is responsible for selecting the optimal voltage vectors in
case that the absolute error between the reference and
predicted values deviates from zero. Then, the cost function
can be expressed by Fig. 4. Rotor flux observer using voltage model with a mismatch in stator
resistance with 20 %
̃ ̃ ,
∆ = (i∗ , − ı̃ , ) ,
+ (i∗ , − ı̃ , ) From Fig.4, it is clear that the effect of Rs variation is
(27) very impressive at low speed operation (low frequency) and
upon this; the controller has to find a solution. One valid

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DE MINAS GERAIS. Downloaded on January 16,2024 at 20:36:27 UTC from IEEE Xplore. Restrictions apply.
305
 2019 IEEE Conference on Power Electronics and Renewable Energy (CPERE)

solution is to adopt a mechanism for estimating the To illustrate the effectiveness of the proposed flux
parameters online, but this will add a computation burden estimator, a bode plot for the transfer function which
and increase the commutation losses. describes the estimator response is presented. The transfer
function which describes the estimator response can be
For that reason, the current paper proposes a procedure expressed after mathematical derivations as follows:
for compensating the error in the estimated rotor flux due to
the parameters variation (here Rs variation is adopted). ∗ ∗ ∗
,
, ,
The proposed estimator takes the measured filtered stator H(S) =
∗
voltage vector u , and stator current vector ı̅ , as inputs
and provides the estimated rotor flux as an output. The (36)
proposed estimator as shown in Fig. 5 consists of two where p refers to the operator d⁄dt , and K is referring to
models; the first one is the current-model(cm) which can be linear state feedback controller (compensator). Then, by
derived directly from the IM model defined in stationary plotting the frequency response of (36), Fig. 6 is obtained
frame by which shows the effectiveness of the proposed flux estimator
and its robustness against the variation in Rs in comparison
ψ = ı̅ , +j ,
ψ (30) with Fig. 4.
, ,

where τ = L ⁄R is the rotor time constant.

For low speed operation, the current-model is assumed to
achieve precise estimation of the rotor flux. From (30), the
stator flux can be evaluated by
ψ , = ψ , + L ı̅ , (31)

The second model in the estimator is the voltage-model(vm)

which estimates the stator flux in the stationary frame using
the following formulation,
ψ = u − R ı̅ −u (32) Fig. 6. Rotor flux observer using proposed flux estimator with a mismatch in
, , , ,
stator resistance with 20 %
It can be noticed that an additional term ( u , ) is The next step is to estimate the rotational speed with the
present in (32) which is used to compensate the errors which help of the estimated rotor flux using the model reference
are generated from a mismatch in Rs and from the pure adaptive regulator (MRAR) which extracts the speed through
integration as well. The compensated term can be evaluated regulating the error between two signals for the rotor flux,
by the help of a PI controller as follows: one of them comes from the proposed estimator shown in
Fig. 5 (it is also entitled as reference model) and the other
u , = k + ∗ ψ , −ψ , (33) signal is generated from a model which is speed dependent
The k and k in (33) can be determined via analyzing the role described by (30) in stationary frame and then can be
of each model inside the estimator such that at zero operating
expressed by
frequency; only the current model (31) action is present; while at ,
high frequencies the voltage model (32) dominates. Based on this, ψ , = ı̅ , +j ψ , (37)
the parameters are calculated as follows:
The error signal e in Fig. 7 can be calculated by:
k = ω1 + ω2 ,and k = ω1 ∗ ω2 (34)
where ω = (2 − 5 ) rad/s and ω = (20 − 30 ) rad/s . =ψ , ψ , −ψ , ψ , (38)
The values ω and ω are very suitable for flexible
conversion from the current-model to voltage model to cover
wide speed range. Finally, the output of the estimator is the us,k ψ̂ r,estk
rotor flux defined in stationary frame by

ψ = ψ − L ı̅ , (35)
, ,
is, k ω̂ me, k
us, k
is, k
ψ̂ cmr, k
est
i s, k
jθψr,k
i sdq, k ψrdq,k jθψr,k
ψ cm
r, k ψ cm
s, k
ucomp,k ψ s, k
e e

Fig. 7. MRAR Observer for rotor speed estimation

θψr,k
ψr,estk
arctg(ψ qr, k ψ dr, k ) The complete system configuration for the proposed PCC
can now be constructed and shown through in Fig. 8.
Fig. 5. Proposed stator and rotor fluxes estimator

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DE MINAS GERAIS. Downloaded on January 16,2024 at 20:36:27 UTC from IEEE Xplore. Restrictions apply.
306
 2019 IEEE Conference on Power Electronics and Renewable Energy (CPERE)

u (i) , i  0,....7 an indication of the torque change; as the torque is increased

u s, k  1 there will be an increase in the torque angle.
Udc

1
u s, k
Z

*
i βs, k 1 i *αs, k 1
i s, k u s, k
is,* k 1 θ*i,k 1

 Fig. 9. a. Rotor Speed (RPM)

ψ r,estk
ω*i,k 1

ωme, k  1 Z 1

ψr,estk  ψr,estk 1
ω*sl, k 1 ΔTs
Ts
ω̂me,k Z 1

Γ12,k 1 est
δ*k 1 δˆ k 1 ψ r, k 1 ω̂ me,k  ω̂ me,k 1
Γ22,k 1 Ts
Fig. 9. b. Torque (Nm)
Γ21,k1 Ts
ω*me, k 1 ω̂me,k 1

Γ*21,k 1
ω̂me,k 1
~
iαs, k  1
~
iβs, k  1
*
Γ12, k 1
Γ *22,k 1

Fig. 9. c. (q-axis) component of stator current (A)

Fig. 8. Proposed PCC control approach

The control procedure as illustrated in Fig. 8 starts with

measuring of stator voltage and current vectors after that the
stator and rotor fluxes are estimated using the proposed flux
estimator illustrated in Fig. 5. Then, the estimated rotor flux Fig. 9. d. (d-axis) component of stator current (A)
is utilized to estimate the angular speed of the motor which is
used to close the speed loop and for prediction of stator
current α-β components at instant (k+1)Ts. The stator current
components with the rotor flux components are used to
estimate the state variables Γ , , Γ , and Γ , using
which the actual torque angle δ is estimated. Then, the
control starts to assign the reference of the stator current Fig. 9. e. Squared rotor flux modulus (Vs)
vector ı̅∗, using the references of the state variables
Γ∗ , , Γ∗ , and Γ ∗ , as given by (24). In the same
manner the reference of the torque angle δ∗ is calculated
by (25). After that the control compares δ∗ with δ
through the torque angle controller and gets the reference
angle of the stator current vector Ѳ∗, which is then used to
Fig. 9. f. Airgap energy (Joule)
get the reference of the stator current components i∗ ,
and i∗ , . The current references are fed with the predicted
currents ı̃ , and ı̃ , to the cost function (27) which
gives the optimal voltages to be applied to the IM.
V. TESTING RESULTS
The IM drive performance is tested for wide range of Fig. 9. g. Torque angle (rad)
operation. The tests start firstly with imposing high speed
Fig. 9. IM dynamic performance at high speed operation
command of 1000 RPM. To test the transient dynamic
response of the drive, a load torque with 2 Nm is applied
since starting till time t=3 s, then the load is increased to the For the low speed operation, the IM drive performance is
full load of 10 Nm. It can be shown through the obtained tested for a speed reference of 30 RPM (1 % of rated speed).
results that a full decoupled has been achieved between the d The rest of test conditions are similar to that of high speed.
and q components of the stator current which are precisely The obtained results shown in Fig. 10 confirm the
tracking their references even at transient operation (increase effectiveness of the proposed speed and flux estimator in
of the torque at t=3 s). In addition, the squared rotor flux preserving both high tracking ability for the reference speed
modulus (Γ , ) is effectively follows its reference value of and for the robustness against the Rs variation. In addition,
0.6 (Vs)2. In addition, the estimated speed ( Γ , ) is the full decoupled between the electromagnetic and
precisely tracking the actual measured speed which assures mechanical components of stator current is achieved even at
the validity of the proposed speed estimator. Finally, the transient change in the load from 2 Nm (light load) to 10 Nm
waveforms of the energy exchange (Γ , ) and estimated (rated load).
torque angle (δ ) are illustrated. The latter is considered as

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DE MINAS GERAIS. Downloaded on January 16,2024 at 20:36:27 UTC from IEEE Xplore. Restrictions apply.
307
 2019 IEEE Conference on Power Electronics and Renewable Energy (CPERE)

the feasibility of the proposed sensorless PCC, the drive

performance is tested for wide range of speed change. The
obtained results confirm the ability of the proposed controller
in obtaining a robust dynamic response from the IM drive
under different operating conditions.

Fig. 10. a. Rotor speed (RPM) APPENDIX A.

TABLE I - PARAMETERS OF (IM) DRIVE
Parameters Value Parameters Value
Rated torque 10 Nm Lm 0.17447 H
Rs 1.50 Ω Kp, KI (Speed Controller) 14.26 and 1263
Rr 0.85 Ω Kp, KI (MRAR observer) 8000 and 65000
Ls 0.1785 H Kp, KI (Energy Controller) 6 and 165
Lr 0.18451 H Kp, KI (Torque angle Controller) 13.5 and 1196
Fig. 10. b. Torque (Nm)

VII. REFERENCES
[1] D. L. Mon-Nzongo, T. Jin, G. Ekemb and L. Bitjoka, "Decoupling
Network of Field-Oriented Control in Variable-Frequency Drives,"
in IEEE Transactions on Industrial Electronics, vol. 64, no. 7, pp.
5746-5750, July 2017.
[2] M. A. Mossa and S. Bolognani, ''Effective model predictive
Fig. 10. c. (q-axis) component of stator current (A) instantaneous power control for a sensorless induction motor
drive'',Int. J. Industrial Electronics and Drives, Vol.4, No. 1, pp.44–
55, 2018.
[3] L. Zhao, J. Huang, J. Chen and M. Ye, "A Parallel Speed and Rotor
Time Constant Identification Scheme for Indirect Field Oriented
Induction Motor Drives," in IEEE Transactions on Power
Electronics, vol. 31, no. 9, pp. 6494-6503, Sept. 2016.
[4] D. Stojic and S. Vukosavic, "Sensorless Induction Motor Drive
Based on Flux Acceleration Torque Control," in IEEE Transactions
Fig. 10. d. (d-axis) component of stator current (A) on Industrial Electronics, vol. 54, no. 3, pp. 1796-1800, June 2007.
[5] G. Brando, A. Dannier, A. Del Pizzo, R. Rizzo and I. Spina,
"Torque derivative control in induction motor drives supplied by
multilevel inverters," in IET Power Electronics, vol. 9, no. 11, pp.
2249-2261, 2016.
[6] M. A. Mossa and S. Bolognani, "Effective model predictive current
control for a sensorless IM drive," 2017 IEEE International
Symposium on Sensorless Control for Electrical Drives (SLED),
Fig. 10. e. Squared rotor flux modulus (Vs) Catania, 2017, pp. 37-42.
[7] J. Scoltock, T. Geyer and U. K. Madawala, "A Comparison of
Model Predictive Control Schemes for MV Induction Motor
Drives," in IEEE Transactions on Industrial Informatics, vol. 9, no.
2, pp. 909-919, May 2013.
[8] Y. Zhang, H. Yang and B. Xia, "Model-Predictive Control of
Induction Motor Drives: Torque Control Versus Flux Control," in
IEEE Transactions on Industry Applications, vol. 52, no. 5, pp.
4050-4060, Sept.-Oct. 2016.
Fig. 10. f. Airgap energy (Joule) [9] M. A. Mossa, " Effective Predictive Flux Control for an Induction
Motor Drive with an Online Estimation Procedure for Stator
Transient Inductance," International Review on Modeling and
Simulations (IREMOS), 2018, Praise Wise Publisher, Vol
11(6):366-376.
[10] Y. B. Zbede, S. M. Gadoue and D. J. Atkinson, "Model Predictive
MRAS Estimator for Sensorless Induction Motor Drives," in IEEE
Transactions on Industrial Electronics, vol. 63, no. 6, pp. 3511-
Fig. 10. g. Torque angle (rad) 3521, June 2016.
[11] M. S. Zaky, "Stability Analysis of Speed and Stator Resistance
Fig. 10. IM dynamic performance at low speed operation Estimators for Sensorless Induction Motor Drives," in IEEE
Transactions on Industrial Electronics, vol. 59, no. 2, pp. 858-870,
Feb. 2012.
VI. CONCLUSION [12] O. Barambones, A. J. Garrido and F. J. Maseda, "Integral sliding-
mode controller for induction motor based on field-oriented control
The paper has presented an effective sensorless predictive theory," in IET Control Theory & Applications, vol. 1, no. 3, pp.
current control (PCC) approach for IM drive. The base 786-794, May 2007.
operation of the proposed PCC is relied on regulating the [13] F. Alonge, F. D'Ippolito and A. Sferlazza, "Sensorless Control of
torque angle of the motor. A set of state variables are utilized Induction-Motor Drive Based on Robust Kalman Filter and
to construct the dynamic model of the IM instead of using Adaptive Speed Estimation," in IEEE Transactions on Industrial
Electronics, vol. 61, no. 3, pp. 1444-1453, March 2014.
classic fifth order model which contributes effectively in
[14] M. A. Mossa and S. Bolognani, "A new formulation of model
reducing the inherent coupling between the electromagnetic predictive direct torque control for a sensorless IM drive," 2017
and mechanical variables of the IM. An effective flux and Nineteenth International Middle East Power Systems Conference
speed estimator is proposed to enhance the robustness of the (MEPCON), Cairo, 2017, pp. 664-670.
controller against uncertainties in the system. To illustrate

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DE MINAS GERAIS. Downloaded on January 16,2024 at 20:36:27 UTC from IEEE Xplore. Restrictions apply.
308