1

A Differential-Algebraic Approach to Speed Estimation in an Induction Motor: Open Loop Experimental Results
Mengwei Li, John Chiasson, Marc Bodson, and Leon M. Tolbert

Abstract Previous work by the authors has described a differential-algebraic approach to speed estimation in induction motors. The method shows that the speed can be found by solving for the roots of a polynomial in whose coefcients are functions of the stator voltages, stator currents, and their derivatives. Preliminary experimental results are presented in which the speed is estimated using the differential-algebraic method. Index Terms Sensorless Speed Observer, Induction Motor

I. I NTRODUCTION Sensorless control of an induction motor refers to the problem of controlling it without the use of a rotor position/speed sensor. Many different techniques have been proposed to estimate the speed of an induction motor without a speed sensor. This area has a rather large literature, and the reader is referred to [1][2][3][4][5][6][7][8] for an exposition of many of the existing approaches. The approach presented in this work is most closely related to the ideas described in [9][10][11][12][13]. In [9][10][11][12], observability is characterized as being able to reconstruct the unknown state variables as rational functions of the inputs, outputs, and their derivatives (See [10][11][12] for a more precise denition). The authors have shown in [14] that the speed can be found by solving for the roots of a polynomial in whose coefcients are functions of the stator voltages, stator currents, and their derivatives. Simulations have shown that this speed estimate can be used in a eld-oriented controller to obtain tracking control of the machine for low-speed (including zero speed) trajectories under full load. However, it does require a small time step size (1 sec) to be able to reconstruct the derivatives of the stator voltages and the stator currents. The control of the machine in real time requires a sample period of 120 sec with the computing platform used by the authors. As a consequence, we consider only the reconstruction of the speed from the stator voltages and stator currents. This is an
M. Li and J. Chiasson are with the ECE Department, The University of Tennessee, Knoxville, TN 37996, mwl@utk.edu, chiasson@utk.edu M. Bodson is with the ECE Department, The University of Utah, Salt Lake City, UT 84112, bodson@ee.utah.edu. L. M. Tolbert is with the ECE Department, The University of Tennessee, Knoxville, TN 37996, tolbert@utk.edu and The Oak Ridge National Laboratory, Oak Ridge TN, tolbertlm@ornl.gov. Drs. Chiasson and Tolbert would like to thank Oak Ridge National Laboratory for partially supporting this work through the UT/Battelle contract no. 4000007596. Dr. Tolbert would also like to thank the National Science Foundation for partially supporting this work through contract NSF ECS0093884.

open-loop procedure in that the estimated speed is not fed back for real-time control of the machine. More specically, the work here describes preliminary experimental results in which the speed is estimated using the algorithm given in Li et al [14], but the estimate is not used for feedback control. The rest of the paper is structured as follows. Section II describes the mathematical model of an induction motor, Section III presents the differential-algebraic speed observer, Section IV presents the stabilized speed observer, Section V gives some simulation results, and Section VI gives the experimental results. The last section gives some conclusions and future considerations.

II. M ATHEMATICAL M ODEL OF THE I NDUCTION M OTOR With iS , iSa + jiSb , R , Ra + j Rb , and uS , uSa + juSb , a two-phase equivalent state-space mathematical model of the induction motor in space vector form is given by (see [15]) d 1 (1 jnP TR ) R iS + u iS = dt TR LS S d 1 M (1 jnP TR ) R + i R = dt TR TR S n o np M d L Im iS = R dt JLR J (1) (2) (3)

where, is the position of the rotor, = d/dt, np is the number of pole pairs, iSa , iSb are the (two phase equivalent) stator currents, uSa , uSb are the (two phase equivalent) stator voltages, and Ra , Rb are the (two phase equivalent) rotor ux linkages, RS and RR are the stator and rotor resistances, M is the mutual inductance, LS and LR are the stator and rotor inductances, J is the inertia of the rotor, and L is the load torque. The symbols TR = LR RR =1 = M2 LS LR

M LS LR

RS 1 1 M2 + LS LS TR LR

have been used to simplify the expressions. TR is referred to as the rotor time constant, and is called the total leakage factor.

III. D IFFERENTIAL -A LGEBRAIC S PEED O BSERVER Differentiating (1) gives d d d d2 i = (1 jnP TR ) R jnP R iS dt2 S TR dt dt dt 1 d + (4) u . LS dt S Using the complex-valued equations (1) and (2), one can d eliminate R and R from (4) to obtain dt 1 d 1 d2 i = (1 jnP TR ) u i + iS dt2 S TR dt S LS S d 1 d M + 2 (1 jnP TR ) iS iS + u TR dt LS dt S jnP TR 1 d d uS iS + iS . 1 jnP TR dt LS dt (5) Solving (5) for d/dt gives 1 jnP TR (1 jnP TR ) d + = 2 dt jnP TR jnP TR d 1 d d2 M (1 jnP TR ) iS iS + uS 2 iS 2 TR dt LS dt dt . (6) 1 d u i + iS dt S LS S If the signals are measured exactly and the dynamic model is correct, the right-hand side must be real. From (1) it is seen that the denominator in the last term of (6) is equal to (/TR ) (1 jnP TR ) R and thus (6) is singular (i.e., the denominator in (6) is zero), if and only if R 0. Breaking down the right-hand side of (6) into its real and imaginary parts, the real part has the form d = a2 (uSa , uSb , iSa , iSb ) 2 + a1 (uSa , uSb , iSa , iSb ) dt (7) + a0 (uSa , uSb , iSa , iSb ) . The expressions for a2 (uSa , uSb , iSa , iSb ) , a1 (uSa , uSb , iSa , iSb ) , and a0 (uSa , uSb , iSa , iSb ) are lengthy and therefore not explicitly presented here (Appendix VIII-B gives their steadystate expressions). It is shown in [14] that (7) is never stable in steady state and thus cannot be used as an observer by just integrating it in real time. On the other hand, the imaginary part of (6) has no derivatives in the speed leading to a second-degree polynomial equation in of the form q() , q2 (uSa , uSb , iSa , iSb ) 2 + q1 (uSa , uSb , iSa , iSb ) + q0 (uSa , uSb , iSa , iSb ). (8) If is the speed of the motor, then q() is zero. The expressions for q2 (uSa , uSb , iSa , iSb ), q1 (uSa , uSb , iSa , iSb ) , and q0 (uSa , uSb , iSa , iSb ) are lengthy and not explicitly presented here (Their steady-state expressions are given in Appendix VIII-A.). There are two solutions to (8), and at least one of these two solutions must track the motor speed. This equation does not have a stability issue, but a procedure is required
2

to determine which of the two solutions is correct. Further, there are situations where the speed cannot be determined by (8). For example, if uSa = constant and uSb = 0, it turns out that q2 (uSa , uSb , iSa , iSb ) = q1 (uSa , uSb , iSa , iSb ) = q0 (uSa , uSb , iSa , iSb ) 0 and is not determinable from (8)1 . On the other hand, if the machine is operated at zero speed ( 0) with a load on it, then q2 (uSa , uSb , iSa , iSb ) 0 and q1 (uSa , uSb , iSa , iSb ) 6= 0, and a unique solution is specied by (8) (see Appendix VIII-A where this is proved in steady state). In fact, for low speed trajectories, consider equation (8) written in the form (q2 + q1 ) + q0 = 0. At low speeds, dened by |q2 | |q1 | , (9) reduces to q1 + q0 = 0 and is uniquely determined by = q0 /q1 . (It is shown in 2 [14] that in steady state, |q2 | |q1 | if (TR np ) 1.) If q2 (uSa , uSb , iSa , iSb ) 6= 0, one determines the correct solution of (8) as follows: Differentiate equation (8) to obtain the new independent equation d (10) + q2 2 + q1 + q0 0. dt Next, d/dt is replaced by the right-hand side of equation (7) to obtain a new algebraic polynomial equation in given by (2q2 + q1 ) g() , 2q2 a2 3 + (2q2 a1 + q1 a2 + q2 ) 2 + (2q2 a0 + q1 a1 + q1 ) + q1 a0 + q0 . (11) (9)

g() is a third-order polynomial equation in for which the speed of the motor is one of its roots. Dividing2 (11) by q() from (8), the polynomial (11) has the form g() = (2q2 a2 + 2q2 a1 q2 q1 a2 + q2 ) q() + r1 (uSa , uSb , iSa , iSb ) + r0 (uSa , uSb , iSa , iSb ) . (12) where
2 r1 (uSa , uSb , iSa , iSb ) , 2q2 a0 q2 q1 a1 + q2 q1 2 2q2 q0 a2 + q1 a2 q1 q2 (13)

and

If is equal to the speed of the motor, then both g() = 0 and q() = 0, and one obtains

r0 (uSa , uSb , iSa , iSb ) , q2 q1 a0 + q2 q0 2q2 q0 a1 + q0 q1 a2 q0 q2 . (14)

r() , r1 (uSa , uSb , iSa , iSb ) + r0 (uSa , uSb , iSa , iSb ) = 0. (15) This is now a rst-order polynomial equation in with a unique solution as long as r1 (the coefcient of ) is
1 An induction machine is not typically operated under these conditions. See [13] for more discussion of this issue. 2 Given the polynomials g(), q() in with deg{g()} = ng , deg{q()} = nq , the Euclidean division algorithm ensures that there are polynomials (), r() such that g() = ()q() + r() and deg{r()} deg{q()} 1 = nq 1. Consequently, if 0 is a zero of both g() and q(), then it must also be a zero of r().

nonzero (It is shown in [14] that r1 6= 0 in steady state if q2 6= 0). The coefcients of r1 , r0 contain 3rd derivatives of the stator currents and 2nd derivatives of the stator voltages and, therefore, noise is a concern. Rather than use this purely algebraic estimator, it is now shown how to combine it with the dynamic model to obtain a smoother (yet stable) speed estimator. IV. S TABLE DYNAMIC S PEED O BSERVER Dividing the right side of the differential equation model (7) by q() (q2 (uSa , uSb , iSa , iSb ) 6= 0) , one obtains where and a2 2 + a1 + a0 = q() + + , a1 a2 q1 /q2 , a0 a2 q0 /q2 .

where , q0 /q1 r0 /r1 if |q2 | 0.05 |q1 | [See (8)] if |q2 | > 0.05 |q1 | [See (15)].

In Appendix VIII-A it is shown that in steady state q1 6= 0 if q2 = 0, while in Appendix ?? it is shown that r1 6= 0 if q2 6= 0. V. S IMULATION R ESULTS As a rst look at the viability of the observer (22), simulations were conducted to test it. Here, a three-phase (twophase equivalent) induction motor model was simulated using S IMULINK with parameter values chosen to be np = 2, RS = 5.12 ohms, RR = 2.23 ohms, LS = LR = 0.2919 H, M = 0.2768 H, J = 0.0021 kg-m2 , L_rated = 2.0337 N-m, Imax = 2.77 A, Vmax = 230 V. Figure 1 shows a block diagram of the speed sensorless control system. The induction motor model for the simulation is based on equations (1), (2), and (3). In the control scheme, the estimated speed is fed back to a current command eldoriented controller [8].
ref +

(16) (17) (18)

d = (t) + (t) (19) dt which is a linear rst-order time-varying system. With (t, t0 ) , e
t t0

Then, as q() 0, (7) may be rewritten as

( )d

Consequently, a sufcient condition for stability is that (t) < 0 for some > 0. It is shown in Appendix VIII-B that > 0 in steady state, so the system is never stable in steady state. For the case that q2 (uSa , uSb , iSa , iSb ) 6= 0, consider (19) to be the induction motor model and the solution of algebraic estimator (15) to be the measurement. Then, let an observer be dened by d = (t) + (t) + ( ) . (20) dt If (t) > > 0 for all t, then the estimator (20) is stable with a rate of decay of the error no less than . As this estimator is the result of integrating the signals (t), (t), and (from (15)), it is a smoother estimate than the purely algebraic estimate. In the case where q2 (uSa , uSb , iSa , iSb ) = 0, then the right side of equation (7) can be divided by q1 (uSa , uSb , iSa , iSb ) +q0 (uSa , uSb , iSa , iSb ) = 0 to obtain d = c(t) + ( ) . (21) dt If > > 0 for all t, then (21) is stable with a rate of decay of the error no less than . The estimate of speed proposed here is dened as the solution to the observer d , a2 (uSa , uSb , iSa , iSb ) 2 + a1 (uSa , uSb , iSa , iSb ) dt (22) + a0 (uSa , uSb , iSa , iSb ) + ( )

the fundamental solution of (19), the full solution is given by (see [16]) Z t (t) = (t, t0 )(0) + (t, )( )d .
t0

Speed Control

iqr +

iq

idr +
id

Current Control

ud uq

uSa
Inverse DQ Transform

uSb

2 / 3
Transform

u S1 uS 2 uS 3

INV

DQ Transform

iSa iSb

3 / 2
Transform

Field Angle

IM

Speed Observer

Fig. 1.

Sensorless speed observer control system.

Figure 2 shows the simulation results of the motor speed and speed estimator with the motor under full load. From t = 0 to t = 0.4 seconds, a constant uSa is applied to the motor to build up the ux and the motor is considered to be held with a brake so that 0. At t = 0.4 seconds, the brake is released and the machine is running on a low speed trajectory ( max = 5 rad/s) with full load on the machine from the start. The estimated speed is used in the eld oriented controller as shown in Figure 1. In this simulation, the observer gain in equation (22) was chosen to be 1000. Figure 3 shows the simulation results of the motor speed and speed estimator with the motor under full load. From t = 0 to t = 0.4 seconds, a constant uSa is applied to the motor to build up the ux and the motor is considered to be held with a brake so that 0. At t = 0.4 seconds, the brake is released and the machine is controlled to a zero speed trajectory ( 0) with full load on the machine. is used in the eld-oriented controller as shown in Figure 1. The observer gain in equation (22) was again chosen to be 1000.

6 4 2
Speed in rad/s

0 -2 -4 -6

Figure 4 shows the experimental results of the motor speed and speed estimator with the motor under full load. From t = 0 to t = 0.4 seconds, uSa = 10 V and uSb = 0 is applied to the motor to build up the ux. At t = 0.4 seconds, the eldoriented controller was used to control the machine running on a low-speed trajectory ( max = 5 rad/s) with full load.
6 5 4 3
Speed in rad/s

-8

2 1 0 -1 -2 -3 0 0.5 1 1.5 2 Time in seconds 2.5 3 3.5

4 Time in seconds

Fig. 2. and with the motor tracking a low speed trajectory (max = 5 rad/s) with full load on the machine from the start.

0 -1
Speed in rad/s

Fig. 4. Motor speed and estimated speed with the motor tracking a low speed trajectory (max = 5 rad/s) with full load from the start.

-2 -3 -4 -5 -6

Figure 5 shows the experimental results of the motor speed and speed estimator with the motor under full load.
2 1 0
Speed in rad/s
0 1 2 3 Time in seconds 4

-1 -2 -3 -4 -5

Fig. 3. and with the motor tracking a zero speed trajectory ( 0) with full load on the machine from the start.

VI. E XPERIMENTAL R ESULTS A three-phase 0.5 hp induction motor was used for the experiment. This machine has the same parameters used in the simulations above. A 10 kHz PWM inverter was used to drive the induction motor and a 4096 pulse/rev optical encoder was attached to the motor for position measurements. A 0.5 hp DC motor was coupled to the induction motor as a mechanical load. The real-time computing system RTLAB from O PAL RT with a fully integrated hardware and software system was used to provide the control signal [17]. Both a low speed and a zero speed trajectory were run. In both experiments, the motor speed was fed back to the eld-oriented controller rather than . The stator voltages and currents were collected and sampled at 120 s. A 3rd order Butterworth lter with cutoff frequency of 30 Hz was used to lter the measured stator voltages and currents. The speed observer (22) was used to obtain the estimated speed . Simulations indicate a signicant improvement in tracking if faster sampling rates are possible.

0.5

1.5

2.5

3.5

Time in seconds

Fig. 5. Motor speed and estimated speed with the motor tracking a zero speed trajectory ( 0) with full load from the start.

From t = 0 to t = 0.4 seconds, uSa = 10 V and uSb = 0 is applied to the motor to build up the ux. At t = 0.4 seconds, the eld-oriented control was used to control the machine running at a zero speed trajectory ( 0) with full load. VII. C ONCLUSIONS This paper introduced a new approach to speed sensorless control of an induction motor by using an algebraic estimate of the speed to stabilize a dynamic speed observer. This observer is stable, does not require any sort of slowly varying

speed assumption, and works at zero speed. The singularities of the observer (i.e., when the leading coefcients of the polynomials in are zero) were characterized under steadystate conditions. The method used 3rd order derivatives of the currents and 2nd order derivatives of the voltages which in turn requires a fast sample rate to accurately compute them. Experimental results of the speed estimator were presented under open-loop conditions. VIII. A PPENDIX : S TEADY-S TATE E XPRESSIONS In the following, S denotes the stator frequency and S denotes the normalized slip dened by S , ( S np ) / S . With uSa + juSb = U S ejS t and iSa + jiSb = I S ejS t , it is shown in [8] under steady-state conditions that the complex phasors U S and I S are related by US ! IS = S 1 + j Sp RS + j S LS S 1 + j Sp = where Sp , RS +
RR S LR (1)S 2 LS TR S 2 1+S 2 2 TR S

Here Sp , IS =

RR S LR

1 S TR

so that i

RS + j S LS RS +

Further,

US h

1+jS S TR 1+jS S TR

(1)S 2 LS TR S 2 1+S 2 2 TR S

US 2 S LS (1+S 2 2 TR ) S +j 1+S 2 2 T 2
S R

uSa iSa + uSb iSb = Re (U S I ) S uSb iSa uSa iSb = Im (U S I ) S and u2 + u2 = |U S |2 Sa Sb As a result, the 1st , 2nd , and 4th terms of q2 are all zero, i.e., !2 d i2 + i2 1 Sa Sb 2 =0 n2 LS TR p 4 dt
2 n2 TR p

i2 + i2 = |I S |2 Sa Sb

S TR . 1

US 2 S LS (1+S 2 2 TR ) S +j 1+S 2 2 T 2
S R

A. Steady-state expressions for q2 , q1 , and q0 The steady-state expressions for q2 , q1 , and q0 are now derived. These expressions are then used to show that q2 > 0 for 6= 0, q2 0 for = 0, and q1 6= 0 if q2 0. The explicit expression for q2 is !2 d i2 + i2 1 Sa Sb 2 2 q2 , np LS TR 4 dt 2 2 2 d iSa + iSb TR (uSa iSa + uSb iSb ) dt 2 2 T uSa + u2 + R i2 + i2 Sa Sb Sb LS 2 2 2 1 M 2 d iSa + iSb + 2 LS TR + TR 4 dt 2 diSa diSb 2 + LS TR iSb iSa dt dt diSa diSb 2 + 2TR iSb iSa (uSb iSa uSa iSb ) dt dt 2 M 2 + LS TR i2 + i2 Sa Sb T R 2 M 2 2 + 2 TR iSa + iSb (uSa iSa + uSb iSb ) . TR In steady state, let (see [8]) uSa + juSb = U S ejS t iSa + jiSb = I S ejS t . The complex phasors U S and I S are related by US . IS = S 1+j Sp RS + j S LS 1+j S
Sp

and n2 p

d i2 + i2 Sa Sb (uSa iSa + uSb iSb ) = 0 dt

2 2 2 1 M 2 d iSa + iSb + 2 LS TR = 0. TR 4 dt

The 3rd term of q2 is given by n2 p

2 2 TR 2 T2 iSa + i2 uSa + u2 = n2 R |I S |2 |U S |2 . Sb Sb p LS LS

The 5th term of q2 is given by 2 diSa diSb 2 n2 LS TR iSb iSa p dt dt = n2 p RS +

The 6th term of q2 is diSa diSb 2 n2 2TR iSb iSa (uSb iSa uSa iSb ) p dt dt 2 S LS (1+S 2 2 TR ) S 2 2TR S |I S |2 |U S |2 2 2 T 2 1+S S R = n2 2 2 L2 1+S2 2 T 2 2 . p ( ) (1)S 2 LS TR + S S 2 2 2S 2R RS + 1+S 2 S T 2 2 S R (1+S S TR ) The 7th term of q2 is 2 M 2 + LS TR i2 + i2 n2 p Sa Sb TR 2 Rs 2 2 2 + (1)Rs TR |I S | |U S | Ls TR 2 = np 2 2 L2 1+S2 2 T 2 2 . ( ) (1)S 2 L T RS + 1+S 2 S TS R + S S 2 2 2S 2R 2 2 S R (1+S S TR )

2 LS TR 2 |I S | |U S | S 2 2 L2 1+S2 2 T 2 2 . ( ) (1)S 2 LS TR S + S S 2 2 2S 2R 2 1+S 2 2 TR S (1+S S TR )

Finally, substituting these steady-state expressions into the expression for q2 , one obtains q2 = RS +
2 n2 TR p (1)S 2 LS TR S 2 1+S 2 2 TR S 2

The 8th term of q2 is M 2 2 np 2 TR i2 + i2 (uSa iSa + uSb iSb ) Sa Sb TR 2 2 2R 1 2 Ls + TR TR |I S | |U S | s =n2 2 2 L2 1+S 2 2 T 2 2 p ( ) (1)S2 L T RS + 1+S 2 S TS R + S S 2 2 2S 2R 2 2 S R (1+S S TR ) (1 ) S 2 LS TR S RS + . 2 1 + S 2 2 TR S

and a0 = where den = np TR |U S | 2 RS +

4
2 2 (1) 1+S 2 TR S 2 TR S S 2 TR 1+S 2 2 TR S

3 (1 ) (1 S) 1 S . 2 2 (1 + S 2 2 TR ) den S

|U S |4 2 2 L2 1+S 2 2 T 2 2 2 ( ) (1)S 2 L T RS + 1+S 2 S TS R + S S 2 2 2S 2R 2 2 S R (1+S S TR )

2 2

(28)

2 LS (1 ) (1 S) S . 2 (1 + S 2 2 TR ) S

|U S | +

4
2 2 1+S 2 2 TR S 2 2 1+S 2 2 TR S

2 L2 S S

2 (23)

(1)S 2 LS TR S 2 1+S 2 2 TR S

2 2 L2 (1+S 2 2 TR ) S S S 2 (1+S 2 2 TR )2 S

(1) S 2 1+S 2 2 TR S
2

R EFERENCES
[1] K. Rajashekara, A. Kawamura, and K. Matsuse, eds., Sensorless Control of AC Motor Drives - Speed and Position Sensorless Operation. IEEE Press, 1996. [2] P. Vas, Sensorless Vector Control and Direct Torque Control. Oxford University Press, 1998. [3] M. Vlez-Reyes, K. Minami, and G. Verghese, Recursive speed and parameter estimation for induction machines, in Proceedings of the IEEE Industry Applications Conference, pp. 607611, 1989. San Diego, California. [4] M. Vlez-Reyes, W. L. Fung, and J. E. Ramos-Torres, Developing robust algorithms for speed and parameter estimation in induction machines, in Proceedings of the IEEE Conference on Decision and Control, pp. 22232228, 2001. Orlando, Florida. [5] M. Vlez-Reyes, Decomposed algorithms for parameter estimation. PhD thesis, Massachusetts Institute of Technology, 1992. [6] M. Vlez-Reyes and G. Verghese, Decomposed algorithms for speed and parameter estimation in induction machines, in Proceedings of the IFAC Nonlinear Control Systems Design Symposium, pp. 156161, 1992. Bordeaux, France. [7] M. Bodson and J. Chiasson, A comparison of sensorless speed estimation methods for induction motor control, in Proceedings of the 2002 American Control Conference, pp. 30763081, May 2002. Anchorage, AK. [8] W. Leonhard, Control of Electrical Drives. 3rd Edition, Springer-Verlag, Berlin, 2001. [9] D. Nei , I. M. Y. Mareels, S. T. Glad, and M. Jirstrand, Software for c control system analysis and design: symbol manipulation, in Encyclopedia of Electrical Engineering, John Wiley & Sons, J. Webster, Editor, 2001. available online at http://www.interscience.wiley.com:83/eeee/. [10] M. Diop and M. Fliess, On nonlinear observability, in Proceedings of the 1st European Control Conference, pp. 152157, Herms, Paris, 1991. [11] M. Diop and M. Fliess, Nonlinear observability, identiability and persistent trajectories, in Proceedings of the 36th Conference on Decision and Control, pp. 714719, Brighton England, 1991. [12] M. Fliess and H. Sra-Ramirez, Control via state estimation of some nonlinear systems, in Symposium on Nonlinear Control Systems (NOLCOS-2004), September 2004. Stuttgart, Germany. [13] S. Ibarra-Rojas, J. Moreno, and G. Espinosa-Prez, Global observability analysis of sensorless induction motors, Automatica, vol. 40, pp. 1079 1085, 2004. [14] M. Li, J. Chiasson, M. Bodson, and L. M. Tolbert, Observability of speed in an induction motor from stator currents and voltages, in IEEE Conference on Decision and Control, pp. 34383443, December 2005. Seville Spain. [15] J. Chiasson, Modeling and High-Performance Control of Electric Machines. John Wiley & Sons, 2005. [16] W. J. Rugh, Linear System Theory. 2nd Edition, Prentice-Hall, 1996. [17] Opal-RT Technologies, RT-LAB, See http://www.opal-rt.com.

With 6= 0, it is seen that q2 > 0 and q2 = 0 if and only if S = 1 (which is equivalent to = 0) . Similarly, it can be shown that the steady-state expression for q1 is q1 = RS + np S |U S |4 2 2 L2 1+S 2 2 T 2 2 2 ( ) (1)S 2 LS TR S + S S 2 2 2S 2R 2 1+S 2 2 TR S (1+S S TR ) 2 2 2 LS (1 ) 1 2 TR (1 S) S . (24) 2 (1 + S 2 2 TR ) S

If = 0, then S = 1 and q1 6= 0. Finally, the steady-state expression for q0 is |U S | 2 (1)S 2 LS TR S + 2 2 T 2 1+S

S R

4
2 2 L2 (1+S 2 2 TR ) S S S 2 (1+S 2 2 TR )2 S 2

q0 = RS +

2 LS (1 ) (1 S) S . 2 (1 + S 2 2 TR ) S

2 (25)

B. Steady-state expressions for a2 , a1 , and a0 The steady-state expressions for a2 , a1 , a0 are n2 |U S |4 p a2 = 2 2 L2 1+S 2 2 T 2 2 2 2L T ( ) (1)S RS + 1+S2 S TS R + S S 2 2 2S 2R 2 2 S R (1+S S TR ) S (1 ) 1 2 2 (1 + S 2 2 TR ) den S
2

(26)

a1 = RS +

2 2 (1 )2 (1 S) 1 S 2 2 (1 + S 2 2 TR ) den S

np |U S | 2 2 L2 1+S 2 2 T 2 2 2 ( ) (1)S 2 LS TR S + S S 2 2 2S 2R 2 1+S 2 2 TR S (1+S S TR ) (27)