Dynamic Control of Torque in Overmodulation and in the ... - IEEE Xplore

IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 21, NO. 4, JULY 2006

1091

Dynamic Control of Torque in Overmodulation and in the Field Weakening Region Anshuman Tripathi, Student Member, IEEE, Ashwin M. Khambadkone, Senior Member, IEEE, and Sanjib K. Panda, Senior Member, IEEE

Abstract—At high angular velocity, the induction motor is operated in the field weakening range due to the voltage limit of the inverter. Field oriented vector control (FOC) is unsuitable for this operation due to coupling, non-linearities, and saturation of linear current controllers. A proposed direct torque control space vector modulation (DTC–SVM) scheme using SVM does not use coordinate transforms or current controllers to achieve DTC. Control of the stator flux vector allows for dynamic change in the torque in all regions, including field weakening with the six-step operation. This paper describes the torque control dynamic in the field weakening range, using a step change in stator flux vector magnitude and it’s angular velocity. The method is demonstrated experimentally. Index Terms—Constant power speed range (CPSR), direct torque control (DTC), field-weakening.

current control in a region where no such decoupling exists. Hence, it would be prudent to achieve dynamic torque control (DTC) in field weakening region by exploiting the inherent coupling. Thus, there two problems, a) How to achieve fast torque dynamic from the overmodulation region into six-step and b) how to achieve good torque dynamic in field weakening. The FOC uses inner current control loop to achieve fast torque dynamics. This is made possible because of the decoupled control structure of the rotor flux oriented coordinate system, where current determines the rotor flux magnitude, and current determines the torque when rotor flux is held constant. The two and components of voltage. The decontrollers produce coupled vector control is valid only when the inverter is able to produce a voltage vector that satisfies

I. INTRODUCTION ONSTANT power operation at high speeds is achieved through field weakening. The constant power region of operation is very useful in traction, electric vehicles, and machine-tool spindle drives. In automotive applications, a high constant power speed range (CPSR) is often used. In addition, a full bus utilization, linear torque response, and for induction motors, six-step operation during transient is desired [1]. In most of induction machines, field weakening happens naturally, when the voltage limit of the drive is reached and any further increase in the stator frequency cannot maintain ratio. The field oriented vector control (FOC) achieves the field weakening by reducing the magnetizing component of stator current , a method similar to the separately excited dc machine. In steady state, the voltage supplied by the inverter needs to compensate for the rotational emf, which is very large ( 1 p.u.). This voltage is supplied by the pulsewidth modulation (PWM) inverter either by operating at maximum modulation index in PWM mode which is equal to 0.906, or by the six-step operation. Current control becomes extremely difficult in overmodulation and six-step due to the presence of low frequency harmonics. A method of compensated current control [2] can overcome the problem. However, the dynamic performance during field weakening is still based on the decoupled current control, where the torque dynamic is achieved by while reducing the increasing the torque producing current component . In so doing, we are trying to force a decoupled

C

Manuscript received January 7, 2005; revised June 10, 2005. Recommended by Associate Editor J. Ojo. The authors are with the Department of Electrical and Computer Engineering, National University of Singapore, Singapore, 117576 (e-mail: eleamk@nus. edu.sg). Digital Object Identifier 10.1109/TPEL.2006.876823

(1) is However, this voltage can only be produced, provided below the maximum possible voltage. Moreover, to satisfy the decoupled structure, the inverter should be able to produce the and components during the dynamic. correct value of The space vector modulation (SVM) scheme, in the normal . The range, produces a voltage that is only 90% of region between is called overmodulation [3]. Steady state and dynamic operation in this region has been presented by many researchers [4]–[7] etc. However, each one of these has some disadvantages and do not provide the best possible response under the given condition. The second problem of torque control in field weakening region using FOC has been investigated in past by Leonhard [8], Joetten [9], Kerkman [10] and others. Depenbrock [11] presented a field weakening scheme using Direct Self Control (DSC). Most of the FOC based schemes rely on external loops or function generators to modify the flux reference. On the other hand, Depenbrock uses rotor flux based slip frequency estimation to determine the path of stator flux for field weakening. Thus, there is parameter dependence and computational overhead in all these schemes. In this paper, we propose a scheme for dynamic control of torque in the overmodulation and field weakening region. The scheme is based on DTC-SVM that has been proposed earlier by many researchers [12]–[16]. These SVM-based torque control schemes achieve constant switching frequency operation unlike the first versions of DTC schemes [11], [17]. In Section II. we will briefly present a variant of DTC–SVM scheme. We will use some of the definitions used in the basic scheme to explain the dynamic overmodulation. We will discuss the torque dynamics in Section III. In Section IV, we address the first problem

0885-8993/$20.00 © 2006 IEEE

1092

IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 21, NO. 4, JULY 2006

Fig. 1. Block diagram of the DFC scheme.

of torque dynamics in overmodulation. A new method for dynamic overmodulation is proposed here. The second problem: DTC in field-weakening is dealt with in Section V. We discuss the problem of torque control in field weakening and propose a simple algorithm for its control . Experimental results of solutions to both problems of dynamic overmodulation and field weakening are presented in Section VI. II. DTC-SVM USING PREDICTIVE STATOR FLUX CONTROL Fig. 1 shows the block diagram of the overall control scheme. The speed controller processes the error between the and the measured speed to generate the commanded speed . In the field weakening range, the referreference torque ence torque is adjusted corresponding to the operating speed to maintain constant power operation, [18]. The torque controller and the synchronous defines the rotor angular frequency . The inner loop angular velocity is obtained as consists of a predictive stator flux vector controller with stator based SVM [15], [19]. The predictive flux error vector flux control and the PWM are different from the other methods is defined by of DTC-SVM. Reference stator flux vector a magnitude and the synchronous angular velocity . The control scheme does not have any current control and is free from any coordinate transforms. For operation of the drive at rotor angular velocities above base value, the magnitude of the rule, where stator flux vector is weakened as per the is the rated magnitude of the flux vector. For low speeds and steady state operation, closed loop stator flux control is achieved by compensating the stator flux error . The volt–seconds required to compensate vector this flux error vector are obtained by switching the inverter switching states similar to space vector modulation Fig. 2. Hence, the on-times for the modulation can be obtained using (2) and are the active states that constitute the sector where is the zero switching state. However, to compensate for and the stator resistance drop, we use (3) is the current vector and is the resistance of the where folstator winding. While the reference flux error vector lows a circular trajectory, switching states govern the trajectory . Switching state times and are obtained of

Fig. 2. Selection of switching state vectors for steady state torque and flux vector control: steady state condition implies,! (k + 1) = ! (k ).

using the principle of volt–second balance as in conventional and angle of the vector SVM using the magnitude of within a sector [see Fig. 2(b)]. During a torque dynamic, the stator flux reference vector will be displaced from the actual stator flux vector by a large angle. This will produce a large stator flux error vector larger than the one shown in Fig. 2(a). However, the maximum magnitude of the flux error vector that can be achieved in a 3 , where 3 , is the sub-cycle cannot be more than normalized maximum value of voltage for a switching state. 3 , it defines the If the flux error vector is greater than transient condition and triggers the large signal algorithm. The is the one switching state nearest to the error vector that can bring about the maximum change . This state is the maximum switching state and is continuously switched. As a result, the fastest possible dynamic response similar to DSC [11], is achieved. The status of the maximum switching state is updated after every sampling period. The stator flux and torque control in normal and overmodulation has been described in our papers [15], [16]. III. TORQUE DYNAMICS DURING DYNAMIC OVERMODULATION In overmodulation, when FOC is used, the voltage limit starts affecting the inner current and torque control loops. A large back emf vector reduces the voltage reserve. Hence, fast increase in torque becomes difficult. Jul et al. [6] and Mochikawa [7] propose a method of overmodulation for FOC, whereas Habetler [13] proposes the direct toque control using space vector modulation (DTC-SVM). The main objective of these methods is to improve the dynamic torque response. All these methods report the application of overmodulation switching only under dynamic operating conditions and therefore are also called dynamic overmodulation methods [20]. During a dynamic condition in the overmodulation region, required for torque control is outside the the voltage vector hexagonal voltage limit of the inverter, Fig. 3. As the inverter cannot provide this voltage, the switching strategy approximates it with a vector that lies on the hexagonal boundary. Different methods of obtaining this approximation have been proposed. Habetler, proposes a DTC based scheme [13], where the approximated voltage vector has no phase difference between the and the readjusted reference voltage required voltage vector vector. Such a vector is shown as in Fig. 3. However, the computation of reference voltage is itself intensive and has division by sinusoidal variables.

TRIPATHI et al.: DYNAMIC CONTROL OF TORQUE IN OVERMODULATION

1093

Fig. 4. Analysis of different switching strategies during a torque dynamic in the overmodulation region.

Fig. 3. Switching strategies for dynamic toque control in the overmodulation region.

On the other hand, the methods proposed in [7] and [6] use the FOC structure and are computationally intensive approaches [20]. In [7], the approximated voltage vector is such that, the error between the magnitude of this vector and that of the reis minimum. This is shown as vector quired voltage vector in Fig. 3. In [6], a vector is formed at the intersection point of a line joining the back-emf vector and the current controller output vector with the hexagon (see Fig. 3). The vector formed by connecting the origin with the intersection point is the readjusted voltage vector given as in Fig. 3. The improvement in the DTC performance of these methods [6], [7] is marginal (as reported in [20]) when compared with the method of [13]. The reason for the marginal improvement lies in the difference in the way torque control is carried out. For instance, methods [7] and [6] have been implemented in a FOC scheme that necessitates the selection of voltage vector such that the field orientation condition is always met and the saturation of current controllers is prevented. On the other hand, the dynamic overmodulation method of [13] has been implemented in a DTC–SVM scheme that relies on fastest possible change in the flux vector during a torque dynamic. However, Habetler’s scheme has few shortcomings, first a phase error will exist as the control is carried out in stator coordinates without prediction. And second, the scheme does not exploit the complete overmodulation region to six-step operation. In Habetler’s scheme, the readjusted voltage vector is achieved using PWM. Since lies on the hexagon, no zero vector is selected, in that sense it is overmodulation. But is this the best strategy to achieve fast torque response? Does PWM created by switching between two active vector produce the fastest response?

IV. DYNAMIC OVERMODULATION TO PRODUCE FAST TORQUE RESPONSE Any voltage vector on the hexagonal limit, is obtained by and as a suitable combination of the voltage vectors, shown in Fig. 3. Fig. 5 compares the effect of selecting different switching options for the dynamic conditions depicted in the Fig. 4. Here, vector is the voltage vector used in [13]. To study

Fig. 5. Current dynamics for different voltage vectors under dynamic conditions.

the effect of switching these vectors, the stator voltage equation will be used. This equation can be arranged as follows:

(4) is the induced-emf vector. Vector can either be where or , or it can be an average vector reswitching state sulting from PWM switching between and . A – coordinate system can be considered, as in Fig. 5, such that the -axis . This is aligned with the stator flux vector. Hence, gives (5) The following observations can be made from Fig. 4, Fig. 5, and (5). • Switching of the voltage vector creates a positive -axis component and positive -axis component of the voltage vector. This increases the positive d-axis current component resulting in an increase in the stator flux vector magnitude (see Fig. 5). As a consequence the magnitude of increases, reducing the -axis current and slowing down the dynamic torque response. • Looking from the stationary coordinate system, the rate of change of torque is proportional to the rate of change of angle between the stator flux vector and the stator current vector. As a result of switching the voltage vector ,

1094

IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 21, NO. 4, JULY 2006

Fig. 6. Study of different switching strategies for dynamic overmodulation: (a) when vector u is continuously switched and (b) when vector u is continuously switched.

Fig. 5, the current magnitude and the stator flux vector magnitude increases. However, the angle between the stator current vector and the stator flux vector reduces resulting in a slower torque dynamic [see Fig. 6(a)]. or is On the other hand, when the voltage vector switched, the following observations can be made using (5), Fig. 4, and Fig. 5. • The extent of dynamic field weakening for the switching is larger than when the voltage vector of voltage vector is applied. This is because of a larger magnitude of the negative -axis component of the current vector. As a rereduces, causing an sult, the magnitude of the vector increase in the -axis component of the current vector. • From the vector geometry of Fig. 4, we can see that when is switched, angle between the stator current vector and the stator flux vector increases at a faster rate than is switched. Hence, switching vector that when vector gives a rapid dynamic torque response as shown in Fig. 6(b). To conclude, the effect of applying the voltage vectors and is different at very high angular velocity. Continuous produces a fast torque dynamic. In [13], the apswitching of and plication of vector will result in the switching of both vectors in every sample but this combination of switching, results in a slower dynamic response. This is because, in every switching period, the momentary field weakening achieved due is partly neutralized by the increase in stator flux to vector due to . The same condition holds true for the schemes [6] or [7] where PWM is done during dynamic overmodulation and such by switching between the two voltage vectors or . On that the resulting average voltage vector is along speeds up the dythe other hand, continuous switching of namic response whereas continuous switching of stabilizes the current. A. Dynamic Overmodulation For the proposed method following switching occurs immediately after a torque step reference is applied. Since the required reference stator flux error vector is large, it will have a large magnitude and a large angle. Under these conditions, the exceeds 6 in a sector. Hence, the angle of the vector voltage vector is switched continuously. This results in a rate of rise of torque and currents as shown in Fig. 7. However, as the

Fig. 7. Result with the proposed method of switching.

Fig. 8. Result with the method proposed in [13].

operation shifts to a new sector, the angle has a value which is is less than 6. Under these conditions voltage vector switched and held on. This checks the rate of rise of currents as explained in (4). Hence, instead of PWM between two active vectors, as in schemes [6], [7], and [13], continuous switching of the voltage vectors is proposed here. On the other hand, Fig. 8 shows the effect of applying the voltage vector as proposed in [13] (vector of Fig. 3). To apply such a vector, both and are switched alternatively in every sampling period such that the volt–sec balance produces a vector . Such a switching results in a slower dynamic response when compared to the proposed scheme. From the above analysis it can be said that, if approximations of the reference voltage vector are used as proposed by any of the schemes, [13], [6] or [7], it will result in a voltage vector along the hexagon boundary for any large reference stator flux . To implement such an adjusted voltage refererror vector ence vector, PWM between the active states needs to be carried out. Therefore, when these methods are employed, the acceleration of the stator flux vector will not be the fastest. Hence, the dynamic response time of the methods proposed in these references will be slower compared with the proposed method. is As opposed to that, in the proposed method, vector 6 giving fastest acceleration and when switched when 6 and sector transition has taken place, the vector of in the next sector, thus the previous sector becomes vector decelerating the stator flux vector. Hence, a time optimal like torque response is produced. This solves the problem of fastest possible dynamic response during overmodulation. Let us now look at the second problem.

TRIPATHI et al.: DYNAMIC CONTROL OF TORQUE IN OVERMODULATION

1095

Fig. 10. (a) Dynamic operation in the field weakening region and (b) switching logic.

Fig. 9. Reference current vector for field oriented current control does not exploit the installed voltage and current capability. Here ! > ! .

V. DYNAMIC TORQUE CONTROL IN THE FIELD-WEAKENING REGION At high angular velocity, the induction motor is operated in field weakening range due to voltage limit of the inverter. In this region of operation, two methods of torque control are prominent a) by Kim et al. in reference [21] that is implemented using FOC and b) by Depenbrock in reference [11] that is implemented in the DSC method of DTC. In [21], the controllable magnitudes of the -axis and -axis current references are obtained based on the available voltage reserve. Instead of weakening the rotor flux vector magnitude rule, the authors use the continusing the conventional uous current rating of the motor and a voltage limit of 0.9 p.u. of the inverter voltage limit to decide the rotor flux magnitude level. The control scheme achieves a better performance than the conventional method of rotor flux vector magnitude weakening. However, steady state current and voltage limits are used for deciding the reference values of the -axis and -axis currents. The current limit used in [21] is expressed as a circular locus while the current limit due to the voltage limit is expressed in the form of an ellipse (see Fig. 9). As the operating angular velocity increases, the size of this ellipse reduces and therefore the controllable -axis and -axis current magnitudes reduce as well. Since the control structure uses motor current limit and 0.9 p.u. of voltage limit, the current and voltage capability of the inverter cannot be exploited under dynamic conditions. The parameter sensitivity of this method was corrected by Harnefors et al. in reference [22]. However, the authors still use the current limit as suggested by the reference [21] and therefore there is no significant improvement in the dynamic response. In a similar approach, Griva et al., [23], use the same voltage limit to generate the torque reference while the reference flux vector magnitude is determined using the – components of the current vector under limiting conditions. Hence, their method requires coordinate transforms. In Depenbrock’s paper, [11], the method of field weakening depends upon the requirement of torque control. The optimum stator flux vector magnitude is calculated based upon the estimated rotor and stator flux vectors and the maximum possible change in the stator flux vector that can be achieved. To this

end, simultaneous equations are solved in every sampling period that require the information of the rotor flux vector. Selection of switching state vectors is done based upon the calculated value of the reference stator flux vector magnitude and the error in torque. Unlike the FOC methods, DSC employs the weakening of the stator flux vector magnitude. This has the advantage of retaining the optimal torque capability [18], [24] during field weakening region. Moreover, it is easier to incorporate in a DTC structure. In the proposed method, we can perform overmodulation into six-steps, hence, in steady state the field-weakening range is carried out in the six-step mode. During the dynamic, the stator flux vector magnitude will be reduced and its angular velocity increased. Hence, the reference stator flux vector travels an inner reference trajectory [see Fig. 10(a)]. However, at that instant, the flux error vector will be very large as shown in Fig. 10(a). For a given rotor angular velocity step, the output of the torque controller produces a step increase in the angular velocity of the reference flux vector. This results in (1) a reduction in the magnitude of the predicted reference vector and (2) a sharp increase of the error vector . The dynamic conin the angle dition is defined when (6) The step change in the angle of the predicted stator reference vector is represented as , where is due to the dynamic. Under such a condition, if the angle of the flux error within a sector, , is greater than 6, vector is continuously switched [see Fig. 10(b)]. The same vector will be switched as long as the dynamic condition and the angle condition exists. 6, When the flux error vector moves into the new sector, will be switched. This has the effect of limhence, vector iting the stator current. The algorithm is the same as the one we used for dynamic overmodulation. Once the flux error has decreased and the actual flux starts following the reference flux, the six-step switching mode occurs naturally. This is because of the switching condition shown in Fig. 10(b). As the synchronous angular velocity is commanded by the torque controller, the appropriate magnitude of the reference stator flux vector during a dynamic in torque is decided by the requirement of torque control. Since no limit is imposed upon the current vector magnitude, the current capability of the inverter is exploited. In Depenbrock’s method [11], the calculation of a) optimum stator flux vector magnitude and b) change in the stator flux

1096

IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 21, NO. 4, JULY 2006

Fig. 12. Control of torque and flux vector magnitude for a step change in the rotor angular velocity from 1.25 p.u. to 2.0 p.u. Fig. 11. Dynamic response for a speed step from 0.85 to 1.35 p.u.: (a) proposed method and (b) method suggested in [13].

vector angle required for fast torque dynamic is based on the slip angular frequency. Calculation of slip angular frequency requires the information of the rotor flux vector which is not readily available in a DSC structure. However, in the proposed method, the slip angular frequency information is given by the torque controller. This avoids the use of rotor parameters which deviate from their normal range values during the field-weakening. Therefore, the proposed method that uses slip and field weakening algorithm which seeks fast reduction in the stator flux vector magnitude, is simple to implement. It exploits the inherent advantages of the DSC structure without needing the parameter dependant calculation of the rotor flux vector. As compared to FOC methods, it does not have problems of controller saturation or low frequency harmonics. Moreover, due to the inherent coupled structure of the induction motor during field-weakening, a decoupled control as in FOC is not very suitable for dynamic field-weakening.

Fig. 13. Flux vector trajectory for a dynamic in the six-step region, rotor angular velocity command 1.3 p.u. to 1.75 p.u.

Fig. 14. Test result showing phase current for a torque dynamic during six-step operation when a speed step from 1.34 p.u. to 1.7 p.u. is given.

VI. EXPERIMENTAL RESULTS The proposed scheme was implemented on a dSPACE/ DS1102 card using a inverter of 2 kW to run a 0.75-kW induction motor. A sampling period of 200 S is used for estimation and control of torque and flux. A modified model based on stator voltage equation is used for flux estimation. It is described in detail in our earlier paper [15]. Estimation of torque is done in block of Fig. 1, using the relationship [25]. Here, is the estimated flux vector while is the measured current vector. The PI controllers for speed and torque are selected using conventional methods. During the sinusoidal modulation and overmodulation range, flux error vector is used to compute the switching states and their on-times as in [15]. During the dynamics a large signal algorithm is used as in [16]. Hence, in the sinusoidal mode the full torque capability of the drive is available. During the field weakening range the stator flux vector magnitude is determined by the look-up table. The angular velocity is determined by the torque controller. For a large dynamic step, the switching state as shown in Fig. 10 is selected. Thus, the scheme can be implemented without an appreciable computational overhead. Fig. 11 gives a comparison of the DTC performance between the proposed method and the method described in [13]. The motor was operating under steady state when a step change in the rotor angular velocity command is given. The dynamic response is much faster in the proposed method due to sharp flux

weakening and rate of rise of currents during the dynamic overmodulation. Note that the currents are within the motor rated current of 1.0 p.u. In the region of operation above base speed, such that the reference torque is limited to a value of constant rated power operation is maintained. Dynamic control of the magnitude of the stator flux vector and torque is shown in Fig. 12. As the phase angle of the error vector at the instant of application of the step is large there is an initial dip in the stator flux vector magnitude. The stator flux vector magnitude is defined by the synchronous angular velocity commanded by the torque controller. Flux vector trajectory for a step increase in the rotor angular velocity command during six-step operation is depicted in Fig. 13. During the transient the stator flux vector magnitude is reduced below its steady state value. This dynamic field weakening causes the current vector magnitude to increase to bring about a fast torque dynamic with six-step operation. In Fig. 14, the magnitude of the current vector for a torque dynamic in the field-weakening region is shown to be within the inverter current limit which is twice the motor rated current. Fig. 15 gives the dynamic torque response in the field weakening range of operation while Fig. 16 shows the inherent current control feature during DTC in the field weakening region. Unlike FOC methods, current control is not done and no deliberate limit is imposed upon the maximum value of phase currents. As voltage limit is already exhausted, the currents rise to

TRIPATHI et al.: DYNAMIC CONTROL OF TORQUE IN OVERMODULATION

1097

REFERENCES

Fig. 15. Test result of torque control for a speed step from 1.34 p.u. to 1.7 p.u.

Fig. 16. Test result showing current with torque dynamic during six-step operation for a speed step from 1.34 p.u. to 2.15 p.u.

Fig. 17. Rotor angular velocity step from 0.1 p.u. to 3.9 p.u.

a value that ensures that the rate of change of active power drawn by the motor is increased during the dynamic period. This implies that the current limit of the inverter is exploited during the dynamics. The result of Fig. 17 shows that smooth control of speed and torque can be achieved in the field-weakening region. In this experiment, a large speed step of 3.8 p.u. was given at an operating angular velocity of 0.1 p.u. This experiment was performed at a voltage that is half the rated dc-link voltage of the inverter so as to maintain the 4 rated speed below the mechanical limit of the loading dc machine. These results demonstrate the ability of the proposed scheme to optimally control the torque by completely exploiting the physical constraints of the inverter. VII. CONCLUSION This paper has shown how steady state and DTC can be done in the overmodulation and the field weakening range without coordinate transforms using the proposed constant switching frequency DTC scheme. Full utilization of the installed voltage and current capability has resulted in a) an excellent DTC performance and b) an extended controllable speed range of approximately 4.0 p.u. of the drive. The scheme is also capable of producing a transient six-step operation. The simple DTC structure makes it suitable for applications like machine tools, traction, and automotive control.

[1] J. M. Miller, S. E. Schulz, B. Conlon, M. Duvall, M. D. Kankam, and N. Nagel, “Adjustable speed drives transportation industry needs part i: Automotive,” in Proc. Veh. Technol. Conf. (VTC’03), Oct. 6–9, 2003, pp. 3220–3225. [2] A. M. Khambadkone and J. Holtz, “Compensated synchronous pi current controller in overmodulation range and six-step operation of spacevector-modulation-based vector-controlled drives,” IEEE Trans. Ind. Electron., vol. 49, no. 3, pp. 574–580, Jun. 2002. [3] J. Holtz, W. Lotzkat, and A. Khambadkone, “On continuous control of pwm inverters in the overmodulation range including the six-step mode,” IEEE Trans. Power Electron., vol. 8, no. 5, pp. 546–553, Oct. 1993. [4] R. J. Kerkman, T. M. Rowan, D. Leggate, and B. J. Seibel, “Control of pwm voltage inverters in pulse dropping range,” IEEE Ind. Appl. Mag., vol. 2, no. 5, pp. 24–31, Sep./Oct. 1996. [5] S. Bolognani and M. Zigliotto, “Novel digital continuous control of svm inverters in the overmodulation range,” IEEE Trans. Ind. Appl., vol. 33, no. 2, pp. 525–530, Mar./Apr. 1997. [6] S. Jul-Ki and S. K. Sul, “A new overmodulation strategy for induction motor drive using space vector pwm,” in Proc. IEEE Appl. Power Electron. Conf., Mar. 1995, pp. 211–216. [7] H. Mochikawa, T. Hirose, and T. Umemoto, “Overmodulation of voltage source pwm inverter,” in Proc. JIEE Int. Soc. Conf., 1991, pp. 466–471. [8] W. Leonhard, Control of Electrical Drives. New York: Springer Verlag, 1985. [9] R. Jotten and G. Maeder, “Control method for good dynamic performance induction motor drives based on current and voltage as measured quantities,” IEEE Trans. Ind. Appl., vol. IA-19, no. 3, pp. 356–363, May/Jun. 1983. [10] T. M. Rowan, R. J. Kerkman, and T. A. Lipo, “Operation of naturally sample current regulators in transition mode,” IEEE Trans. Ind. Appl., vol. IA-23, no. 4, pp. 586–596, Jul./Aug. 1987. [11] M. Depenbrock, “Direct self control(dsc)of inverter-fed induction machines,” IEEE Trans. Power Electron., vol. PE-3, no. 4, pp. 420–429, Oct. 1988. [12] L. Xu and M. Fu, “A sensorless direct torque control technique for permanent magnet synchronous motors,” in Proc. IEEE Ind. Appl. Conf., 1999, vol. 1, pp. 159–164. [13] T. G. Habetler, F. Profumo, M. Pastorelli, and L. M. Tolbert, “Direct torque control of induction machines using space vector modulation,” IEEE Trans. Ind. Appl., vol. 28, no. 5, pp. 1045–1053, Sep./Oct. 1992. [14] M. P. Kazmierkowski and G. Buja, “Review of direct torque control methods for voltage source inverter-fed induction motors,” in Proc. IEEE 29th Annu. Conf. Ind. Electron. Soc. (IECON’03), Nov. 2003, pp. 981–991. [15] A. Tripathi, A. M. Khambadkone, and S. K. Panda, “Stator flux based space vector modulation and closed loop control of the stator flux vector in overmodulation into six-step mode,” IEEE Trans. Power Electron., vol. 19, no. 3, pp. 775–782, May 2004. [16] A. Tripathi, A. M. Khambadkone, and S. K. Panda, “Torque ripple analysis and dynamic performance of a space vector modulation based control method for ac-drives,” IEEE Trans. Power Electron., vol. 20, no. 2, pp. 485–492, Mar. 2005. [17] I. Takahashi and T. Noguchi, “A new quick response and high-efficiency control strategy of an induction motor.,” IEEE Trans. Ind. Appl., vol. IA-22, no. 5, pp. 820–827, Sep./Oct. 1986. [18] R. Krishnan, Electric Motor Drives Modelling, Analysis and Control, 1st ed. Englewood Cliffs, NJ: Prentice-Hall, 2001. [19] A. Tripathi, A. M. Khambadkone, and S. K. Panda, “Space-vector based, constant frequency, direct torque control and dead beat stator flux control of ac machines,” in Proc. IEEE Int. Conf. Ind. Electron., Contr., Instrum. Autom. (IECON’01), Nov. 2001, vol. 2, pp. 1219–1224. [20] A. M. Hava, S.-K. Sul, R. J. Kerkman, and T. A. Lipo, “Dynamic overmodulation characteristics of triangle intersection pwm methods,” IEEE Trans. Ind. Appl., vol. 35, no. 4, pp. 896–907, Jul./Aug. 1999. [21] S. H. Kim and S. K. Sul, “Maximum torque control of an induction machine in the field weakening region,” IEEE Trans. Ind. Appl., vol. 31, no. 4, pp. 787–794, Jul./Aug. 1995.

1098

IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 21, NO. 4, JULY 2006

[22] L. Harnefors, K. Pietiläinen, and L. Gertmar, “Torque-maximizing field-weakening control: Design, analysis, and parameter selection,” IEEE Trans. Ind. Electron., vol. 48, no. 1, pp. 161–168, Feb. 2001. [23] G. Griva, F. Profumo, M. Abrate, A. Tenconi, and D. Berruti, “Wide speed range dtc drive performance with new flux weakening control [for induction motor drives],” in Proc. 29th Annu. IEEE Power Electron. Spec. Conf. (PESC’98), May 1998, vol. 2, pp. 1599–1604. [24] X. Xu and D. W. Novotny, “Selection of the flux reference for induction machine drives in the field weakening region,” IEEE Trans. Ind. Appl., vol. 28, no. 6, pp. 1353–1358, Nov./Dec. 1992. [25] P. K. Kovacs, Transient phenomenon in Electrical Machines. Amsterdam, The Netherlands: Elsevier, 1984.

Anshuman Tripathi (S’04) received the B.Eng. degree (with first class honors) from the Government Engineering College, Rewa, India, in 1999, the M.Tech. degree from the Indian Institute of Technology, Kanpur, in 1999, and the Ph.D. degree from the National University of Singapore in 2006. He is currently with GE R&D, Bangalore, India. His areas of interest include control of electrical drives, power converters, power electronics and machine design.

Ashwin M. Khambadkone (SM’04) received the Dr.-Ing. degree from Wuppertal University, Wuppertal, Germany, in 1995 and the Graduate Certificate in education from the University of Queensland, Brisbane, Australia. At Wuppertal, he was involved in research and industrial projects in the areas of PWM methods, fieldoriented control, parameter identification, and sensorless vector control. From 1995 to 1997, he was a Lecturer at the University of Queensland. He was also at the Indian Institute of Science, Bangalore, India in 1998. Since 1998, he has been an Assistant Professor at the National University of Singapore. His research activities are in the control of ac drives, design and control of power electronic converters, and fuel cell based systems. Dr. Khambadkone received the Outstanding Paper Award in 1991 and the Best Paper Award in 2002 both which appeared in the IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS.

Sanjib K. Panda (SM’00) received the B.Eng. degree from South Gujarat University, Surat, India, in 1983, the M.Tech. degree from the Institute of Technology, Bhu, India, in 1987, and the Ph.D. degree from the University of Cambridge, Cambridge, U.K., in 1991, all in electrical engineering. Since 1992, he has held a faculty position in the Department of Electrical Engineering, National University of Singapore and is currently an Associate Professor. His research interests include control of electrical drives and power electronics based systems. Dr. Panda received the Cambridge-Nehru Scholarship, the M. T. Mayer Graduate Scholarship, and the IEEE Third Millennium Medal. He has served as the Technical Programme Chairman for the IEEE PEDS’97 Conference and is the Organizing Chairman of IEEE PEDS’03 Conference.