Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

18th Mediterranean Conference on Control & Automation

Congress Palace Hotel, Marrakech, Morocco


June 23-25, 2010

Backstepping Control for an Induction Motor with an Adaptive


Backstepping Rotor Flux Observer
Ramzi Trabelsi , Adel Kheder , Med Faouzi Mimouni , Faouzi M’sahli

Abstract—This paper deals with the synthesis of a stabilizing that achieves global asymptotic rotor speed tracking for the
Backstepping nonlinear control provided with an adaptive full-order, nonlinear model of an induction motor despite the
Backstepping observer to estimate the rotor flux of an Induction uncertainty in rotor resistance and time-varying load torque
Machine (IM). This approach is based on the vector control conditions. In [14] the robust adaptive backstepping
strategy by replacing the existing conventional controller with a controller for efficiency optimization of IM with
Backstepping control in order to more quickly bring back the q- uncertainties is designed. The backstepping control is
axis flux towards zero and to reduce the effect of the motor
parameters variations. To observe the rotor flux, we have
applied to induction motor drive with good performance for
introduced an observer based on the Backstepping theory trajectories tracking [2] and [6-7]. However, the performance
provided with an adaptive mechanism of rotor resistance. The accuracy is affected by load torque and the machine
design principle of the controlled system is established using the parameters uncertainties. To overcome this problem, an
Lyapunov stability theory. estimation of the rotor resistance is done in this paper for
compensation of its unknown variations, which constitutes
I. INTRODUCTION one of the contributions of this work. Some researches have
proposed various IM drives with on-line rotor resistance or
I NDUCTION motor (IM) compared to other types of
electric machines, has found to be used in a wide range of
industrial application, due to its excellent reliability great
rotor time-constant adaptation [1], [3] and [5-7] to produce
better control performance, such as a sliding mode observer,
robustness and less maintenance. However, the control of the a model reference adaptive system and Luenberger observer.
induction motor is complex because the dynamics model of In this paper, a new adaptive backstepping observer is
the machine is nonlinear, multivariable highly coupled and as developed for the simultaneous estimation of the rotor flux
well as the state variable are not easily measurable and they components and of the rotor resistance for an IM under the
are dependent on the machine parameters variations [14]. assumption that only the stator currents and the motor speed
One of the most significant developpement in this area has are available for measurement. The controller of IM is
been the filed oriented control proposed by blaschke [4], designed based on backstepping technique. In addition, an
However, this strategy is very sensitive to parameters adaptive law based on the Lyapunov stability theory
variation. To improve filed oriented control, many modified estimates the rotor resistance. The combination of these two
nonlinear state feedback schemes have been proposed such methods represents both a contribution in this work and a
as, sliding mode control, input-output linearization control, good possibility to achieve classical control objectives for
passivity based control, flatness-control, direct torque systems having unmodeled or parasitic dynamics and
control, nonlinear predictive control and backstepping parametric uncertainties.
control. The paper is organized as follows; Section 2 recalls the
In the last two decades, the backstepping technique has dynamic model expressed in the rotor field oriented control
been widely studied and developed for the control and state (d-q) frame. Then the synthesis of the nonlinear
estimation problems since the studies of Kanellakopoulos, Backstepping control is presented. Section 4, deals with the
Kokotovic and Morse [8]. This control technique allows a Backstepping flux observer provided with an adaptation of
good steady state and good dynamic behaviour in the the rotor resistance variations. The stability study based on
presence of system parameters variation and disturbances [3] Lyapunov theory is presented in section 5. Simulation results
and [14-15]. Several methods of applying the backstepping are illustrated and discussed in section6.
control to induction motor drives have been presented. In
[15] the nonlinear adaptive backstepping control is designed II. CONTROL PROBLEM STATEMENT
to compensate the unknown system parameters and
disturbances. In [3] a new adaptive backstepping controller A. Induction motor model
The equivalent tow-phase model of the symmetrical IM,
Manuscript received January 11, 2010. This work was supported in part under assumptions of linear magnetic circuits and balanced
by the Research Unit “Commande Numérique des procédés Industriels operating conditions, is presented in a reference rotating
99/UR/11-29: COMPRI-Gabes-Tunisia”. frame (d-q) [11-12], as
Ramzi Trabelsi, Adel Kheder , Med Faouzi Mimouni , Faouzi M’sahli,
are with the National Engineering School of Monastir, Ibn Eljazzar city,
5019, Monastir Tunisia, phone: 00216 98 384 212; e-mail: trabelsi.ramzi@
yahoo.fr,[mfaouzi.Mimouni, faouzi.msahli]@ enim.rnu.tn).

978-1-4244-8092-0/10/$26.00 ©2010 IEEE 5


d better applied to replace the traditional nonlinear feedback
 dt isd = −γ isd + ωs isq + kα rϕ rd + kωrϕ rq + β usd plus proportional-integral PI control of the field oriented
 control technique to ensure better performances.
 d i = −ω i − γ i − kω ϕ + kα ϕ + β u Backstepping design is divided into various design steps. In
 dt sq s sd sq r rd r rq sq

 each step we essentially deal with an easier, single input-


d single-output design problem, and each step provides a
 ϕ rd = M α r isd − α rϕ rd + ωsl ϕ rq (1)
 dt reference for the next design step. The overall stability and
d performance are achieved by a Lyapunov theory for the
 dt ϕ rq = M α r isq − ωslϕ rd − α rϕ rq whole system [2], [11-12]. The synthesis of this control can
 be achieved in three successive steps.
d np f
 dt ωr = J (Tem − Tl ) − J ωr i/ Step 1
In the first step, it is necessary that the system can follow a
where the state variables are the stator currents ( isd , isq ), given trajectory. It corresponds to make the design of a
the rotor fluxes ( ϕ rd , ϕ rq ) and the electrical rotor speed ωr . controller to ensure a good tracking and regulation. we
define yc = (ωc , ϕc ) , where ωc and ϕc > 0 are speed and flux
The stator voltages ( usd , usq ) and slip frequency ωsl
reference trajectories. The speed tracking error eω and the
isq
( ωsl = ωs − ωr = M α r ) are considered as the control flux magnitude tracking error eφ are defined by:
ϕ rd
variables.  eω = ωc − ωr
 (5)
with the constants defined as eφ = ϕc − ϕrd
M2 M2 M 1 Using Eq. (3), we can write the derivative of Eq. (5) as
Rλ = Rs + 2 Rr , σ = 1 − , µ= , β= ,
Lr Ls Lr Lr σ Ls follows:
1 eω = ωc − f1
k = βµ , γ = β Rλ , α r = .  (6)
Tr eφ = ϕc − f 2
The electromagnetic torque is given by: However, the tracking performances will be satisfactory if
3 np M we choose the first Lyapunov candidate function
Tem = (ϕ rd isq − ϕrq isd ) (2) v1 associated to the flux and speed errors:
2 Lr
eω2 + eφ2
B. Rotor field oriented control v1 = (7)
2
Assuming that de d-axis is held by the rotor flux phasor, Using Eq. (8), the derivative of Eq. (7) is written as follows
the IM equations given by (1) turn to be:
v1 = eω (ωc − f1 ) + eφ (ϕc − f 2 ) (8)
d d d
ω = f1 , ϕ rd = f 2 , isd = f3 + β usd , Therefore, the derivative of the Eq. (7) becomes negative
dt dt dt
(3) definite such as:
d d
isq = f 4 + β usq , θ s = ωr + ωsl v1 = −k1eω2 − k2 eφ2 (9)
dt dt
The different functions defined in the above Eq.(3) are With k1 and k2 are positive parameters.
expressed as: To satisfy the conditions mentioned above, it is necessary to
np f satisfy the following equalities
f1 = kcϕ rd isq − Tl − ωr , f 2 = M α r isd − α rϕ rd , eω = ωc − ωr = −k1eω
J J  (10)
isq2 eφ = ϕc − ϕ rd = −k2 eφ
f 3 = −γ isd + ωr isq + M α r + kα rϕ rd , (4) where,
ϕ rd
isd isq ωr = ωc + k1eω
f 4 = −ωr isd − M α r − γ isq − kωrϕ rd  (11)
ϕ rd ϕ rd = ϕc + k2 eφ
and Eq.(11) permits to generate these reference currents assured
2 the Lyapunov stability condition. These currents are given
3 np M by:
kc =
2 JLr  1
(isd )c = M α (ϕc + α rϕ rd + k2 eφ )
 r
III. SPEED AND FLUX BACKSTEPPING CONTROLLER  (12)
The Backstepping control mechanisms for the rotor (i ) = 1 (ω + n p T + f ω + k e )
 sq c kcϕ rd c J l J r 1 ω
angular speed regulation and the flux generation can be
ii/ Step 2

6
In this step, we try to make isd and isq becomes as the observer via an adaptive law based on a Lyapunov
desired in the previous step. Let us define other errors stability theory [1], [5] and [9-10].
between the currents and their references We consider the stator currents as the system outputs,
y1 = x1 , y2 = x2 (20)
eid = (isd )c − isd
 (13) The BFO can be constructed as;
eiq = (isq )c − isq  xˆ1 = −γˆ xˆ1 + kαˆ r xˆ3 + kωr xˆ4 + β u sα + vα
While substituting Eq. (3) in the derivative of Eq. (13), we 
have:  xˆ2 = −γˆ xˆ2 − kωr xˆ3 + kαˆ r xˆ4 + β usβ + vβ
 (21)
eid = (isd )c − isd = (isd )c − f 3 − β usd  xˆ3 = M αˆ r xˆ1 − αˆ r xˆ3 − ωr xˆ4
 (14) 
eiq = (isq )c − isq = (isq )c − f 4 − β usq  xˆ4 = M αˆ r xˆ2 + ωr xˆ3 − αˆ r xˆ4
Setting Eq. (12) in Eq. (13), we obtain: where,
t t
 1 [ x1 x2 x3 x4 ] = isα is β ϕ rα ϕ r β  (22)
eid = M α (ϕc + α rϕ rd + k2 eφ ) − isd
 r R ∆R
 (15) αˆ r = α r + ∆α r = r + r , Rˆ λ = Rs + M µαˆ r (23)
e = 1 np f Lr Lr
(ω + T + ω + k e ) − isq
 iq kcϕ rd c J l J r 1 ω and
iii/ Step 3 Rˆ R M µ∆α r
Since the real inputs commands of the machine ( usd , usq ) γˆ = λ = λ + = γ + ∆γ (24)
σ Ls σ Ls σ Ls
have appeared in Eq.(14), we can go to the final step. Now, where xˆi is the estimation of xi for i ∈ {1, 2,3, 4} . vα and
we define a new Lyapunov function based on the errors of
the speed, the flux and the stator currents vβ are the control input to be designed by backstepping
eω2 + eφ2 + ei2d + ei2q method. So, all parameters of induction motor will be
v2 = (16) considered as constant except rotor resistance. The rotor
2
resistance Rr is considered as uncertain parameter with Rrn
By setting Eq. (14) in the derivative Eq. (16), one obtain:
as its nominal value.
v2 = −k1eω2 − k2 eφ2 − k3 ei2d − k4 ei2q + eid (k3 eid + (isq )c − f 4 − β usq )
We define state error equation as follows:
+ eiq (k4 eiq + (isd )c − f3 − β usd )  ey1 = ex1   x1 − xˆ1 
ei =  =  (25)
(17)  ey2 = ex2   x2 − xˆ2 
In order to make the derivative of the complete Lyapunov
and
function Eq.(17) be negative definite, the quantities between
parentheses in Eq. (17), are chosen equal to zero.  ex3   x3 − xˆ3 
eϕ =   =   (26)
 k3 eid + (isq )c − f 4 − β usq = 0  ex4   x4 − xˆ4 
 (18)
 k4 eiq + (isd )c − f 3 − β usd = 0 where ei and eϕ , the stator current and the rotor flux
The stator voltages controls are given by: observer errors, respectively.
Using Eqs.(20), (21), (25) and (26), the dynamical equations
usd = (k4 eiq + (isd )c − f 3 ) / β
 (19) for the prediction errors are
usq = (k3 eid + (isq )c − f 4 ) / β ex = −γ ex + k (α r ex + ωr ex ) + ∆α r ( Mx1 − xˆ3 ) − vα
( )
 1 1 3 4

Where k3 and k4 are positive parameters selected to


ex2 = −γ ex2 + k (−ωr ex3 + α r ex4 ) + ∆α r ( Mx2 − xˆ4 ) − vβ
( )
ensure that the dynamics current converge is faster than those 
of the flux and speed. ex3 = −(α r ex3 + ωr ex4 ) + ∆α r ( Mx1 − xˆ3 )

ex4 = (−ωr ex3 + α r ex4 ) + ∆α r ( Mx2 − xˆ4 )
IV. ADAPTIVE BACKSTEPPING OBSERVER
(27)
The synthesis of the Backstepping control considers that In the case where the machine operates without rotor
the state variables are available for feedback and the IM resistance variations α r = αˆ r et γ = γˆ ( ∆α r = 0 ), the
parameters are known exactly. Unfortunately, the rotor
Eq.(27) takes the following form:
fluxes are inaccessible for measurement. In this section we
will propose a new algorithm based on Backstepping ex1 = k (α r ex3 + ωr ex4 ) − vα

observer under the assumption that only the stator currents ex2 = k (−ωr ex3 + α r ex4 ) − vβ
and the rotor speed are available for measurement. The rotor  (28)
fluxes and rotor resistance are replaced by those delivered by ex3 = −(α r ex3 + ωr ex4 )
e = −(−ω e + α e )
 x4 r x3 r x4

7
In order to solve the tracking problem, and to overcome the So that the Lypunov function is defined positive and its
difficulties that arise due to the unavailability of one of the derivative definite negative it is necessary that:
state variables for measurement [11-12], a nonlinear tracking k k x3ex3
controller is proposed using the observer based backstepping ∆Rr ( zα x3 − k µ zα x1 + z β x4 − k µ z β x2 −
Lr Lr Lr
algorithm. (36)
i/ Step 1 x4 ex4 1 d ∆Rr
+ µ x1ex3 − + µ x2 ex4 + )=0
The first step in the backstepping observer, is to design a Lr δ r dt
stable controller for the integral of the prediction errors which leads to write the mechanism of adaptation of the rotor
ey1 , ey2 using ex1 , ex2 as virtual control variables with resistance as follows:
stabilizing functions ( ycα , ycβ ), where ycα and ycβ are d ∆Rr x3ex3 x4 ex4
= δ r (k µ z β x2 + − µ x1ex3 + − µ x2 ex4
reference trajectories for virtual variables. dt Lr Lr
(37)
then we choose the integral of the currents prediction errors k k
exα , exβ as follows − zα x3 + k µ zα x1 − z β x4 )
Lr Lr
exα = ex1
 (29)
exβ = ex2
and we define the tracking errors zα and z β of the tracking
prediction currents errors ey1 , ey2 , by
 zα = e y1 − ycα
 (30)
 zβ = ey2 − ycβ
Knowing that the observer's dynamics is based on the
exponential convergence, we can impose the variations of the
reference trajectories for virtual variables as follows
 yc1 = −λ1exα
 (31)
 yc2 = −λ1exβ
ii/ Step 2
In this step, setting the Eq.(31) in Eq.(30), we can find the
Fig.1. Block diagram of the designed drive system
expression of the tracking errors zα and z β
 zα = e y1 + λ1exα VI. SIMULATION RESULTS
 (32)
 zβ = ey2 + λ1exβ We present in this section the simulation results of the
proposed control strategy applied to the two horse power
Using Eqs.(27) and (29) and Eq.(32) and selecting following
squirrel cage IM. The characteristics of the IM are
control inputs
summarised in appendix1. The different parameters
vα = k (α r ex3 + ωr ex4 ) + λ1ex1 + λ2 zα + exα k1 , k2 , k3 and k4 of the backstepping control are chosen as
 (33)
vβ = k (−ωr ex3 + α r ex4 ) + λ1ex2 + λ2 zβ + exβ follows: k1 = 1800, k2 = 200, k3 = 500, k4 = 200 . The
By setting Eq. (33) in Eq. (32), one obtain: developed control law suppose that only the stator current
 zα = −γ ex1 + k ∆α r ( Mx1 − xˆ3 ) − λ2 zα − exα and the mechanical speed are available. We have made some
 (34) scenarios of functioning in order to characterise and improve
 zβ = −γ ex2 + k ∆α r ( Mx2 − xˆ4 ) − λ2 z β − exβ the performances of the command structure.
where λ1 and λ2 are positive design constants that 1) From the first scenario, we have illustrated the response
determine the closed loop dynamics. of the machine under trapezoidal form reference speed; a
very good speed and flux regulation are obtained. The
V. ADAPTIVE LAW OF ROTOR RESISTANCE AND STABILITY corresponding results are illustrated by figure2.
ANALYSIS 2) A second scenario consists to test the sensitivity of the
To design the observer and parameter updating laws, one can control structure performances to change on the rotor
define the following Lyapunov candidate. resistance trajectory with and without adaptation. From the
steady state regime related to the first scenario, we introduce,
1 R2 
Ve =  e2x + e 2x + zα2 + z β2 + ex23 + ex24 + r  (35) at the time t=0.6s, a load torque equal to 4 N.m and
2  α β δ r  respectively at t=1s and t=3s we have introduced an increase
Where δ r is positive design constant of adaptive gain. of rotor resistance equal to 50% and 100%. The

8
corresponding results are illustrated by figure3. [3] E. Arbin and M. Gregory, Adaptive Backstepping Control of a Speed-
Sensorless Induction Motor Under Time-Varying Load Torque and
3) By figure 4, we have illustrated in the same time, the Rotor Resistance Uncertainty. Proceedings of the 38th Southeastern
simulation results curves for three functioning mode; without Symposium on System Theory, Tennessee Technological University
parameters variations, variation without adaptation of rotor Cookeville, TN, USA, March 5-7, 2006, pp 512-518.
resistance and variation with adaptation of rotor resistance. [4] F. Blaschke, The principle of field orientation applied to the new
transvector closed loop system for rotating field machines, Siemens
The results given by figure 2 are obtained if the Rev., 34, 1972, pp. 217-220.
parameters of the machine are supposed to be constant. This [5] H. Nademi, F. Tahami and M. Rezaei, Fault Tolerant IPMS Motor
simulation is carried out by applying the reference speed Drive Based on Adaptive Backstepping Observer with Unknown
Stator Resistance, 2008 3rd IEEE Conference on Industrial
illustrated by figure 2 (a), the real speed converges perfectly Electronics and Applications, ICIEA 2008, 2008, p 1785-1790.
to their reference. By figure 3, we present the machine [6] H. Tan and J. Chang, "Adaptive Backstepping control of induction
response with variation of rotor resistance without and with motor with uncertainties," in Proc. the American control conference,
San Diego California, June 1999, pp. 1-5.
adaptation. For that one considers variations of the rotor [7] H.T. Lee, L.C. Fu, and F.L. Lian, "Sensorless adaptive backstepping
resistance illustrated by figure3 (b). By figure 4, one presents speed control of induction motor," In Proc IEEE Conference on
a whole of the comparative recordings which show the Decision Control, USA, December 2002, pp. 1252-1257.
[8] I. Kanellakopoulos, P. V. Kokotovic and A. S. Morse, Systematic
improvement of the performances of the suggested control
Design of Adaptive Controllers for Feedback Linearizable Systems,
brought by the online rotor resistance adaptation. IEEE Trans. Automat. Contr., vol. 36, 1991, pp. 1241-1253.
[9] J. Soltani, A. Farrokh Payam and M.A. Abbasian, A Speed Sensorless
VII. CONCLUSION Sliding-Mode Controller for Doubly-Fed Induction Machine Drives
with Adaptive Backstepping Observer, Proceedings of the IEEE
In this paper, we have been developed a control law for International Conference on Industrial Technology ICIT, 2006, pp.
induction motor drive by using both Backtepping strategy 2725-2730.
[10] M. Jalalifarl, A. Farokhpayam, S. Sadeghi, S. M. Saghaeiannejad,
and Lyapunov stability theory. The proposed structure Sensorless speed dynamic modeling of an induction motor with input-
contains the Backtepping rotor flux observer associated to output feedback linearization controller and adaptive nonlinear,
online adaptive rotor resistance estimation. This control INMIC2007 - 11th IEEE International Multitopic Conference, 2007.
[11] O. Elmaguiri, F. Giri, Digital backstepping control of induction
carries out asymptotically the output variables process motors, IEEE International Symposium on Industrial Electronics,
towards a reference model established starting from the 2007, pp. 221-226.
desired performances for the closed loop system. The study [12] S. Chaouch , L. Chrifi, A. Makouf, S. Nait, S. Med, Backstepping
of the used observer confirms its robustness in terms of the control analysis of two different speed sensorless approaches for
induction motor, 5th International Multi-Conference on Systems,
parametric variations and external disturbances. The Signals and Devices, SSD'08, 2008.
insertion of an adaptive mechanism of rotor resistance [13] W. Jia-Jun and H. Jie, Torque and flux direct backstepping control of
consolidates of advantage the robustness characters of the induction motor, Proceedings of the World Congress on Intelligent
Control and Automation (WCICA) , June 2008, pp. 6402-6406.
system. [14] Y. Ho Hwang, K. Kwang Park and H. Won Yang, Robust Adaptive
Backstepping Control for Efficiency Optimization of Induction
REFERENCES Motors with Uncertainties, IEEE International Symposium on
Industrial Electronics, 2008, pp. 878-883.
[1] A. Farrokh Payam and B. Mirzaeian Dehkordi, Nonlinear sliding-
mode controller for sensorless speed control of DC servo motor using
[15] Y. Tan, J. Chang and H. Tan, adaptative backstepping
adaptive backstepping observer, 2006 International Conference on control and friction compensation for AC servo with
Power Electronics, Drives and Energy Systems, PEDES '06, 2006. inertia and load uncertainties,IEEE Trans. on industrial
[2] C. Huang, K.L. Chen, H.T. Lee and L.C. Fu, "Nonlinear adaptive electronics, Vol.50, No. 5, October 2003, pp. 944–952.
backstepping motion control of linear induction motor," in Proc. the
American control conference anchorage, May 2002, pp. 3099-3104.

9
150 4
zoom around 0
3.5
100
100 3

magnitude stator current (A)


estimated rotor speed (rd/s)
50
99 2.5

Reference, real and

Real and observer


0 98 2

97 1.5
-50

96 1
0.6 0.61 Estimation
-100
(a) Ref erence 0.5 Observ er
Real Real
-150 0
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
(a) (b)

1 10

0.8
5
Reference, real and observer

Electromagnetic and load


magnitude rotor flux (wb)

0.6
Ref erence

torque (N.m)
Observ er 0
Real
0.4

-5
0.2 Load
Estimated
Real
0 -10
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
(c) (d)
Time (s) Time (s)

Fig.2. Sensitivity of the system performances to change in the reference speed; (a) reference, real and estimated rotor speed; (b) real and observed
magnitude stator current; (c) reference, real and observed magnitude rotor flux (d) electromagnetic and load torque .

120 9

100
8
Reference speed (rd/s)

80 Rrn 1.5*Rrn
7 2*Rrn

ofrotor resistance
Reference signal
60

40 6

20
5
0
4
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 1 1.5 2 2.5 3 3.5 4 4.5 5
(a) (b)

2 25

20 Without adaptation
1.5
Errors rotor speed (%)

Without adaptation of Rr
Errors rotor flux (%)

0.2
0.2
15
1
10 00
With adaptation of Rr
0.5 With adaptation
5 Zoom
-0.2
-0.2 1 2 3 4 5
0 0 1 2 3 4 5

-0.5 -5
1 1.5 2 2.5 3 3.5 4 4.5 5 1 1.5 2 2.5 3 3.5 4 4.5 5
(c) (d)

40 40
Errors stator current (%)

30 30
Without adaptation
Errors torque (%)

Without adaptation
20 20

With adaptation
10 With adaptation 10

0 0

-10 -10
1 1.5 2 2.5 3 3.5 4 4.5 5 1 1.5 2 2.5 3 3.5 4 4.5 5
(e) (f )
Time (s) Time (s)

Fig.3. Sensitivity of the system performances to change on the rotor resistance trajectory without and with adaptation ; (a) reference speed; (b) reference
signal of rotor resistance; (c) errors rotor speed; (d) errors rotor flux; (e) errors stator current; (f) errors torque.

120 10
Reference and estimated of Rr
Reference speed (rd/s)

100
8
80 9.5 Zoom around 3 s
6 9
60 7 Zoom
4 around 1 s 8.5
40 6.5
8
20 2 6 7.5
0 3 3.05 3.1
1 1.05 1.1
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
(a) (b)
Electromagnetic torque (N.m)

100
4
Rotor speed (rd/s)

80 100
w ithout adaptation 4 4
60 2 with adaptation of Rr
99 w ith adaptation 3.9 without adaptation
40 3.9
reference speed without parameters v ariations
0 3.8 3.8
20 98 1 1.05 1.1 3 3.05
1 2 3 4
0
-2
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
(c) (d)

1
Modulus current (A)
Modulus flux (wb)

3
0.8

0.6 2

0.4 w ithou adaptation w ith adaptation


w ith adaptation 1 data2
0.2
w ithout parameters variations data3
0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
(e) (f)
Time (s) Time (s)

Fig.4. Effect of adaptation of rotor resistance; (a) reference speed; (b) reference and estimated rotor resistance; (c) rotor speed; (d) electromagnetic torque;
(e) magnitude rotor flux; (f) magnitude stator current.

10

View publication stats

You might also like