Converter Temperature Regulation with Dual Mode

Converter Temperature Regulation with Dual Mode Control of Fault-Tolerant Permanent Magnet Motors W.U.N Fernando, L Papini and C Gerada School of Electrical and Electronic Engineering The University of Nottingham Nottingham, NG7 2RD, UK. Email: [email protected], [email protected], [email protected]

Abstract—A control strategy for fault tolerant PM motors is developed and addresses the operation during a cooling failure in the power-electronic converter. A fault tolerant PM motor interfaced with a parallel H-bridge per-phase topology is considered. Conventional pulse-width modulation (PWM) based voltage generation is used under normal operation. However, the parallel H-bridge module per-phase topology benefits from the capability to generate quasi-square wave (QSW) voltages as an alternative to PWM voltages. The generation of a QSW voltage incurs lower switching losses compared with PWM based voltage generation and the former can be utilized as a safe operating mode for the converter. Typically, converters designed for aerospace applications are equipped with liquid cooling systems and may also require continuous duty operation under cooling system failure. This paper considers the possibility to alternate between these two switching methods for the regulation of converter temperature under contingency e.g., during reduced cooling conditions. Simulated operating point waveforms of current, voltage and torque during PWM and QSW modes are presented for a 18kW 20krpm machine. A reduction in losses in the range of 20% − 50% is estimated via simulations. Thermal simulations present the temperature regulation capability with the dual mode controller.

I. I NTRODUCTION Mechanical rotary systems have seen an increase in the use of direct-driven and geared electrical machines in many industrial applications. For example, aero-engine starter/generator operation with variable speed electrical machines are presently being researched due to its potential to achieve reduced weight and improved reliability of the mechanical-electrical power conversion stage [1]–[4]. Aero-engine starter operation requires fault-tolerant characteristics and the permanent magnet (PM) machine is found to be a candidate for such applications due to its high power density and the operational capability with fault-tolerance. A typical geared-drive aero-engine starter operation may require motoring capability up to 30krpm [5] and generating capability at even higher speeds. Conventional method of PM motor control involves PWM inverters with high switching frequency and is capable of instantaneous control of the PM motor phase currents [6]–[8]. However, as the PM motor electrical frequency is increased, a higher switching frequency is required to provide the same level of controllability of phase currents. This will increase the power electronic losses in inverters designed with the present generation of high-power semiconductor technology.

978-1-4673-0803-8/12/$31.00 ©2012 IEEE

vdc

Phase - a

i2

Phase - a H bridge

Fig. 1.

Phase - c

i3

i1

Phase -b H bridge

Phase - b

Phase - c H bridge

High speed PM machine

Permanent magnet motor and parallel H-bridge module topology.

Typically, the power converters associated with the starter/generators are designed for operation with liquid cooling [9]. However, the fault-tolerant requirements demand the system operational capability under a cooling failure. One option to accommodate this requirement is to design the converter with air-cooling. However, due to the associated space constraints, such a design option may render infeasible. The investigation of this paper is focused on fault-tolerant PM machines interfaced with the parallel H-bridge per-phase topology as shown in figure 1. In contrast with the conventional star-connected topology, the parallel H-bridge per-phase topology offers the capability to control individual phases with a high degree of independence and provides higher availability under faulted conditions. One additional feature of this topology is the capability to impose zero voltage on a conducting phase for any arbitrary length of time by turn-on/turn-off of the appropriate switches. This capability facilitates the generation of quasi-square wave (QSW) voltages as shown in figure 4. This form of voltage generation can be utilized for the control of PM machines even in high-speed operation, especially for intermittent-duty applications. The circuit topology of figure 1 allows either QSW voltage based operation or PWM based operation and hence is capable of providing smooth transition between either mode. The QSW voltage generation requires turn-on and turn-off operation of the switches twice per-electrical-cycle. Hence, this strategy generates lower power electronic losses in comparison with a high frequency PWM based switching. This feature can be utilized for the operation of the inverters under contingency, such as converter over-temperature which may occur under a

1902

Inverter temperature

where vn and in are the per-phase instantaneous AC side terminal voltage and current. The flux linkage ψn incorporates the effects of both the stator flux and the rotor PM flux. Rn is the stator resistance and lnn represents the cosinusoidal variation of self inductance. The flux-linkage due to the PM is represented in the format of a complex Fourier series,

over-temperature shutdown quasi-square wave mode

over-speed shutdown

PWM switching mode

ψpm,n = − 21

(3)

k=−K1

electrical frequency f

Fig. 2.

K1 o n X ψf,k e−jk(θe +γn )

where ψf,0 = 0 and ψf,k represents the k th harmonic magnitude of the PM flux-linkage. (ψf,1 > 0). The input voltage can be represented in the format of a complex Fourier series:

Dual mode temperature-frequency regions of operation.

converter cooling failure. The rationale of this paper is to formulate a dual mode PM motor controller that regulates the converter temperature by switching between PWM and QSW strategies. Figure 2 shows a temperature-frequency operating region and illustrates the different modes of operation. The per-phase mathematical representation of the faulttolerant PM machine is described in section II. Section III describes the control design for dual-mode operation and seamless transition between the two modes with the converter thermal protection strategy. Section IV present simulation results for a 18kW 20krpm machine. Per-phase current, voltage and total torque production waveforms during three key operating points are presented. The semiconductor losses during nine operating points are compared. Temperature regulation during a 30min cyclic loading operation is simulated and the resultant temperatures in the heat-sink and junctions are presented. II. P ERMANENT MAGNET MOTOR HARMONIC MODEL The system considered in this paper is based on a threephase fault-tolerant PM motor fed by a parallel H-bridge inverter topology. The circuit representation of the inverter and the motor is shown in figure 1. Each phase of the machine is electrically isolated by means of an H-bridge module. The fault-tolerant design of the machine permits the assumption of full magnetic and thermal isolation between the machine phases. Although the harmonics injected by a PWM voltage can be manipulated via the duty cycle, higher-order harmonic components injected by a QSW voltage are a function of the fundamental voltage and are not directly controllable. The torque components generated by the interaction of higher-order currents and non-sinusoidal flux-linkage distribution within a PM machine needs to be considered in order to provide seamless transition between the two modes of control. Hence, a harmonic model [3], [10] that represent the motor per-phase dynamics is used for the control design. The continuous-time dynamics of a general nth phase of a fault-tolerant PM machine can be written as: dψn vn = Rn in + (1) dt ψn = lnn in + ψpm,n (2)

vn (t) =

K X

1 2

{v n,k e−jk(θe +γn ) }

(4)

k=−K

where v n,k = −vd,n,k + jvq,n,k , and v n,−k = v ∗n,k The phase current can be expressed in the form of an exponential series, iph,n =

1 2

K X

{in,k e−jk(θe +γn ) }

(5)

k=−K

where the complex coefficients are defined by, in,k = −id,n,k + jiq,n,k , in,0 = 0 and in,−k = i∗n,k (6) and the current components in-phase and in anti-quadrature phase with the PM flux linkage can be written as: iq,n,k = ipk,n,k cos kεk and id,n,k = ipk,n,k sin kεk

(7)

It follows from [3] that the interaction between the k − 1 and k + 1 harmonics, and the k th harmonic is given by the general complex differential equation: d(in,k−2 + in,k+2 ) din,k La jkωin,k − + Lb (8) dt dt −jkω in,k−2 + in,k+2 − Rs in,k + vdc v n,k − jωkψf,k = 0 Lq −Ld Lq +Ld where La = and La = . The per-phase 2 2 average torque production can be represented as [3]: Tavg,n =

K −kpp X 2kϕf,k in,k 8

(9)

k=−K

+ (Lq − Ld ) i∗n,k+2 − i∗n,k−2 in,k The nth -phase, k th harmonic real power (Pn,k ) and reactive power (Qn,k ) components at the inverter AC side can be written as,

1903

Pn,k =

1 Re v n,k i∗n,k 2

and Qn,k =

1 Im v n,k i∗n,k 2 (10)

Hysteresis controller

Speed controller

T*

PI

_

å

+

+

+

Θ*

wr*

1

å

wr

_

PM machine

0

Θ

MTPA input lookup for THIPWM

un

Switching waveform generator Current waveform generator

wr

d ( t )¢

+

+

å _

d (t )

HIPWM

PI

in

a'

Θ

π − 2α

[α , β ]

{ θon+

θoff+ θon−

{

MTPA Input lookup for quasi-square wave voltage

å

Current controller

in*

wr

+

θoff−

Phase - a H bridge

{

α +β π + β −α

Fig. 3.

Block diagram representation of the speed controller and per-phase mode controller.

phase voltage

{ {

θ on+ α +β

θ off+

back-EMF

π

θ on−

θ off−

2π

θ

electrical angle

{

−vdc

III. MTPA TORQUE AND SPEED CONTROL

π − 2α

+vdc

π + β −α

Fig. 4. Quasi-square wave voltage and back-EMF fundamental waveforms + + − − and the corresponding α, β, θon , θof f , θon and θof f positions.

The control for the PM motor considered in this study is formulated on a maximum torque per-ampere (MTPA) strategy. MTPA schemes are popular for PM motor control and have been utilized in different forms [11]. The MTPA scheme adopted in this paper is based on off-line calculation of the control signals via sequential quadratic programming (SQP) [12]. This generates the required voltage/current commands in PWM mode and the turn-on/turn-off angles for quasi-square wave mode. The MTPA optimization is formulated as: " min

A. mode 1:

{un,k }

In PWM mode of switching, the input voltage command is directly linked with the PWM duty via: v (t) d (t) = vdc

(11)

N X

(

n=1

K X in,k i∗n,k

k=1

)#

2 Ibase

Subject to T∗ −

Nf X

Tavg,n = 0 and dn (t) ≤ 1

π π π and − ≤ β ≤ 2 2 2 for all n and t 0≤α≤

where the v (t) is given by (4). B. mode 2: In quasi-square wave mode, the turn-on and turn-off angle values are linked with the input voltage direct and quadrature harmonic components: 4vdc vn,q = cos (nα) cos (nβ) (12) nπ 4vdc cos (nα) sin (nβ) (13) vn,d = nπ Fundamental real power / torque and reactive power can be manipulated by α and β values. The relationship between α and β, and the fundamental voltage components is given by: r π 2 + v2 α= cos−1 v1,q (14) 1,d 4vdc v1,d β = tan−1 (15) v1,q

for PWM mode or

n=1

for QSW mode

Here Ibase represents the base current value and Nf the number of healthy phases of the considered PM motor. The torque demand is calculated by the speed control loop. The speed loop is implemented as a proportional integral (PI) controller, where the average torque demand is calculated by, Z ∗ ∗ T = kp,s (ωr − ωr ) + ki,s (ωr∗ − ωr )dt (16) A. Mode 1 (PWM based voltage generation): In this mode, for the given torque demand the optimal phase voltage harmonic components are obtained by a multidimensional lookup table that is generated by the MTPA optimization explained earlier. The expected current waveform harmonic components can also be calculated online by solution of equation (8). The PWM switching is then calculated by a

1904

vd,1, vq,1 and corresponding α and β for QSW−mode operation

vd,1, vq,1, vd,3 and vq,3 for HIPWM operation

−1 −2 −1

B. Mode 2 (Quasi-square wave voltage generation):

The dual mode operation for thermal protection is implemented via a hysteresis controller: switch to mode − 2 if Θ∗ − Θ < −h (18) switch to mode − 1 if Θ∗ − Θ > h where Θ∗ and Θ are the desired and actual temperatures respectively. Parameter h represents the hysteresis band of the controller.

1 0 −1 −2 −1

1

vq,1 command [pu]

1 0

0.5 0 −0.5 (e) Torque demand [pu]

1

0.5 0 −0.5 (f) Torque demand [pu]

1

0.5 0 −0.5 (g) Torque demand [pu]

1

−0.5 0 0.5 (b) Torque demand [pu]

1

0.5 −0.5 0 (h) Torque demand [pu]

1

80 60

0

40 20

−0.5 −1

0 −0.5 0.5 (c) Torque demand [pu]

0 −1

1

0.5

100 50

0

−0.5 −1

TABLE I M OTOR DRIVE SYSTEM PARAMETERS

0

−0.5 −1

0.5

IV. S IMULATION RESULTS

1

0.5

−1 −1

vd,3 command [pu]

C. Thermal protection:

−0.5 0 0.5 (a) Torque demand [pu]

2

1.5

vq,3 command [pu]

In this mode, instantaneous current control is not possible. For the given torque demand, the optimal turn-on and turn-off angles are obtained by a separate multi-dimensional lookup table which is also calculated off-line by the same MTPA optimization. Figure 3 shows a block diagram representation of the dual mode controller with mode-1 PWM control strategy and mode2 quasi-square wave voltage generation.

vd,1 command [pu]

0

vq,1 command [pu]

d (t) is a feed-forward duty term calculated by (4) and (11) which corresponds to the desired MTPA terminal voltage.

1

α [deg]

0

2

β [deg]

(17)

vd,1 command [pu]

feedback/feedforward current control law, Z 0 d (t) = d (t) + kp,c (i∗n − in ) + ki,c (i∗n − in )dt

0

−50 0 0.5 −0.5 (d) Torque demand [pu] 4000rpm

1 10000rpm

−100 −1

16000rpm

20000rpm

Fig. 5. d-axis and q-axis voltage components for HIPWM and QSW modes of operation with the MTPA optimization. (a) vd,1 , (b) vq,1 , (c) vd,3 , and (d) vq,3 for HIPWM mode operation, (e) vd,1 , (f) vq,1 for QSW mode operation, and the corresponding (g) α, (h) β in QSW mode.

Parameter

Value Machine parameters: kW rating 18kW PM flux ψf,1 0.0645Vs Maximum speed ωm 20krpm Base speed ωm 10krpm Rs 0.01Ω Lq 1.8mH Ld 1.6mH Pole pair number 2 DC-link voltage 135V Inverter switching frequency 40kHz H-bridge parameters (per-phase): Maximum voltage 600V Maximum current 260A On-state resistance 5.3mΩ (IGBT) 5.0mΩ (Diode) Forward voltage 2.4V (IGBT) 1.5V (Diode) IGBT 10% fall time 500ns (Typical) IGBT tail time 500ns (Typical)

A. MTPA optimization:

A 2-pole 18kW PM machine interfaced with a parallel Hbridge converter topology described by the parameters given in table I has been simulated with the control system shown in figure 3. The MTPA optimization is performed for a torque demand variation T ∗ ∈ [−1pu, 1pu] for the PWM and QSW modes. In the former, harmonic injection PWM (HIPWM) is considered with the fundamental and the third harmonic voltages.

Figure 5 (a) to (d) present the HIPWM mode MTPA optimal variation of the fundamental and the third harmonic d-axis/qaxis voltages for a variable torque demand. Figure 5 (c) and (d) present the QSW mode fundamental d-axis/q-axis voltages for a variable torque demand. The corresponding α and β associated with the QSW fundamental voltages are shown in figure 5 (g) and (h) respectively. By comparison of figures 5 (a), (b) and (e), (f), it can be observed that the QSW mode yields a higher fundamental voltage and hence has a higher DC-link voltage utilization. However, at high speeds, the maximum torque production capability of the QSW mode is limited at a lower value compared with the HIPWM mode. The third harmonic is controllable in HIPWM mode and hence achieves a higher maximum torque capability compared with QSW mode. Figures 5 (g) and (h) reveal that the operation above the base-speed 10000rpm yields α = 0 and torque control is achieved by the variation of β. This represents field-weakening operation. During the operation below the base-speed 10000rpm, the α and β values

1905

v1

2

Operating point 9 - HIPWM mode

0

iph ,1[A]

200

Torque

4

v1

2

2 0

iph ,1[A]

4 (c) Time [ms]

4

−100

6 0

100

100

0

2

4 (d) Time [ms]

6

0

Operating point 9 - QSW mode

Torque

4

2

−100

−100 0

0

[V] and

200

2

6

v1

0

8

1

3

2

0

0

1

2

(e) Time [ms] Phase voltage (switch average in PWM mode)

Phase current

3

Motor torque (Te) [Nm]

4

0

0

Torque

100

[V] and

6

15

10 (b) Time [ms]


200

Torque

5

v1

100

0

0



−100

iph ,1[A]

0

15


10 (a) Time [ms]

8

[V] and

iph ,1[A]

200

5


0

Upper IGBT−1 losses [W]

−100

−100

[V] and

5

Upper IGBT−2 losses [W]

0

10

Lower IGBT−1 losses [W]

iph ,1[A] [V] and

5

100

Torque

v1

iph ,1[A]

0

v1

100

[V] and

10


200

Lower IGBT−2 losses [W]

Torque




200

60 40 20 0

1

2

3 4 5 6 7 8 9 (g) Operating point number

1

2

3 4 5 6 7 8 9 (h) Operating point number

1

2

3 4 5 6 7 8 (i) Operating point number

9

1

2

3 4 5 6 7 8 (j) Operating point number

9

60 40 20 0

60 40 20 0

60 40 20 0

(f) Time [ms]

Harmonic injection PWM mode

Total electromagnetic torque

Quasi-square wave (QSW) mode

Fig. 6. Current, voltage and torque waveforms in PWM and QSW modes for the operating points 3, 6 and 9, and H-bridge IGBT power losses comparison for all the nine operating points given in table II. (a), (c) and (e) PWM mode waveforms and (b), (d) and (f) QSW mode waveforms for operating points 3, 6 and 9. PWM and QSW modes power losses in the (g) upper IGBT-1, (h) upper IGBT-2, (i) lower IGBT-1 and (j) lower IGBT-2.

correspond to the MTPA optimal condition which yield torque from both the PM and reluctance components. B. Operating point evaluation: TABLE II O PERATING POINTS CONSIDERED IN THE SIMULATION AND THE CORRESPONDING CONVERTER LOSSES ( PER H- BRIDGE ) Operat− ing point

Speed [rpm]

Torque [Nm]

1 2 3 4 5 6 7 8 9

4000 4000 4000 10000 10000 10000 16000 16000 16000

5.1 8.5 11.9 5.1 6.8 8.5 1.7 3.4 5.1

Total loss HIPWM mode [W] 107.0 187.7 259.1 110.8 163.8 231.5 61.0 108.0 176.4

Total loss QSW mode [W] 50.8 119.4 182.6 87.2 120.7 154.3 37.5 75.2 117.5

Reduction in losses % 52.5 36.4 29.5 21.3 26.3 33.4 38.5 30.4 33.4

For the evaluation of converter power losses, nine operating

points of interest have been extracted from the simulations. Table II present the parameters associated with these nine operating points and the corresponding power losses during HIPWM and QSW modes of operation. The current, voltage and torque waveforms in PWM and QSW modes associated with the operating points 3, 6 and 9 are shown in figures 6 (a), (c) and (e), and (b), (d) and (f) respectively. The simulated IGBT power losses in one H-bridge module for the nine operating points are also shown in figures 6 (g) to (j). It can be observed from figures 6 (a) to (f) that in each operating point the average torque demand is produced accurately and the controller is operational. The torque production in HIPWM contains a lower ripple magnitude compared with the QSW mode. This can be considered as the main undesirable characteristic in QSW mode. In addition, the high harmonic content of magnetic flux-densities within the machine in QSW mode may also cause high electromagnetic losses and remains to be investigated. Evaluation of the torque quality and the machine losses are out of scope of this paper and will be

1906

70

60

50

40 0

5

10

15 (a) Time [mins]

20

25

30

100

80 70 60 50 40 0

5

10 15 20 25 (b) Time [mins]

30

90 80 70 60 50 40 0

5

Thermal responce with HIPWM operation during cooling failure

10 15 20 25 (d) Time [mins]

30

Lower IGBT Junction temperature 0C

80

90

100

Lower diode Junction temperature 0C

90

Upper IGBT Junction temperature 0C

Heat-sink Temperature 0C

100

100

Upper diode Junction temperature 0C

110

100

90 80 70 60 50 40 0

5

25 10 20 15 (c) Time [mins]

30

5

10 15 20 25 (e) Time [mins]

30

90 80 70 60 50 40 0

Thermal responce with QSW operation during cooling failure Thermal responce with dual-mode temperature regulation during cooling failure

Fig. 7. Thermal simulation results for the H-bridge module and heat-sink. (a) Heat-sink temperature during PWM mode, QSW mode and hysteresis controlled mode, and (b)-(e) IGBT junction temperatures in one H-bridge module.

C. Thermal simulation: The QSW mode loss mitigation capability is considered for the regulation of converter temperature during a cooling failure. To investigate this possibility, a thermal model of the H-bridge converter is simulated with steady-state power loss information. The H-bridge module thermal model is

Simulated drive speed ω and torque demand T* profile Torque demand T* [Nm]

12

16000 Speed ω [rpm]

presented elsewhere. It can be observed from figures 6 (g) to (j) that during all the nine operation points considered, the IGBT losses during PWM mode operation are higher than that in QSW mode operation. The IGBT switching and conduction losses are a function of the commutated current magnitude and the DClink voltage. Hence, high losses can be seen during high torque conditions where a high current is fed to the motor. Figures 6 (g) to (j) also reveal that the relative difference in overall power losses in the upper IGBTs are lower than that of the lower IGBTs during QSW mode operation. This is mainly due the high duty of the lower IGBTs during one cycle, i.e., the upper IGBT current commutation occurs only during the positive and negative voltage pulses while the lower IGBT current commutation occurs during the zero-voltage free-wheeling periods in addition to the positive and negative voltage periods. In QSW mode field-weakening (α = 0) operation, the upper and lower IGBTs have equal duty and incur similar losses. However, the total losses remain lower than that of PWM mode operation. Evaluation of the total losses (as shown in table II) reveals that a higher level of loss mitigation can be expected during low-speed low-load conditions, e.g., 52.5% in operation point 1. At high loads and high speeds, the loss mitigation is in the neighbourhood of 30%.

10

12000

8 6

8000

Torque

Speed

4

4000

2 0

1

2

3

5 4 Time [s]

6

7

8

9

Fig. 8. Cyclic load/speed profile considered in the thermal simulation (one cycle = 9s).

implemented from manufacturer data [13]. Under normal operation, the heat-sink is assumed to be equipped with forced cooling which yields a heat-sink to ambient thermal resistance Rsa < 0.2K/W. During forced cooling failure, the heatsink is assumed to be designed for a heat-sink to ambient thermal resistance Rsa ≈ 0.4K/W. The thermal response during a forced cooling failure is simulated with the PM motor operation with the cyclic load/speed profile shown in figure 8. This profile incorporates the nine operating points presented in table II and is repeated in a 30 minute simulation. Figure 7 (a) present the heat-sink thermal response during PWM and QSW modes and also with the hysteresis controlled mode. The hysteresis controlled mode regulates the temperature at a nominal value of 85 0 C by switching between PWM mode and QSW mode. Figures (b) to (e) present the junction temperatures of the IGBTs in one H-bridge module during the temperature regulated operating condition.

1907

The QSW operation is applied only with intermittent duty due to the hysteresis action. During this mode, the junction temperatures remain within acceptable limits. The concept of temperature regulation at an upper boundary by alternating between the PWM and QSW modes is thus corroborated by this simulation. V. C ONCLUSION Parallel H-bridge per-phase converter topology for faulttolerant PM motors and the converter power losses associated with PWM mode and QSW mode operation have been investigated in this paper. The capability to regulate converter temperature during cooling failure by alternating between these two control strategies have been studied via simulations. A PM motor operation with PWM mode and QSW mode are compared for nine operating points of interest. During all operating conditions, the QSW mode is shown to achieve lower converter losses at the expense of higher torque ripple and possible higher electromagnetic losses. Preliminary simulations show that this loss mitigation in the converter is mainly in the range of 20% to 35% and around 50% during low-load/low-speed conditions. This difference in power loss associated with these two modes of control is utilized for converter temperature regulation during a converter cooling failure. Simulations show successful regulation of temperature by alternating between these two strategies. Future work will consider experimental validation and further evaluation of electromagnetic losses associated with the machine in the QSW mode operation.

[10] P. Chapman, S. Sudhoff, and C. Whitcomb, “Multiple reference frame analysis of non-sinusoidal brushless dc drives,” Energy Conversion, IEEE Transactions on, vol. 14, no. 3, pp. 440 –446, sep 1999. [11] G. Foo and M. Rahman, “Sensorless sliding-mode mtpa control of an ipm synchronous motor drive using a sliding-mode observer and hf signal injection,” Industrial Electronics, IEEE Transactions on, vol. 57, no. 4, pp. 1270 –1278, april 2010. [12] J. N. Nocedal and S. J. Wright., Numerical Optimization. Prentice Hall, 1999. [13] “SKM200GB063D data sheet,” SEMIKRON Elektronik Verwaltungs GmbH, Nrnberg, Deutschland.

R EFERENCES [1] A. Mitcham and J. Cullen, “Permanent magnet generator options for the more electric aircraft,” Power Electronics, Machines and Drives, 2002. International Conference on (Conf. Publ. No. 487), pp. 241–245, June 2002. [2] W. Fernando, M. Barnes, and O. Marjanovic, “Direct drive permanent magnet generator fed ac–dc active rectification and control for moreelectric aircraft engines,” IET Electric Power Applications, vol. 5, no. 1, pp. 14–27, 2011. [3] W. Fernando and M. Barnes, “Model based optimization and fault tolerant control of permanent magnet machines with harmonic injection pulse width modulation,” in Vehicle Power and Propulsion Conference (VPPC), 2011 IEEE, sept. 2011, pp. 1 –6. [4] W. Fernando, M. Barnes, and O. Marjanovic, “Excitation control and voltage regulation of switched reluctance generators above base speed operation,” in Vehicle Power and Propulsion Conference (VPPC), 2011 IEEE, sept. 2011, pp. 1 –6. [5] S. MacMinn and W. Jones, “A very high speed switched-reluctance starter-generator for aircraft engine applications,” in Aerospace and Electronics Conference, 1989. NAECON 1989., Proceedings of the IEEE 1989 National, May 1989, pp. 1758 –1764 vol.4. [6] H. Ransara and U. Madawala, “A new matrix converter based three phase brushless dc motor drive,” in Industrial Electronics (ISIE), 2011 IEEE International Symposium on, june 2011, pp. 89 –94. [7] H. Ransara and U. Madawala, “A low cost brushless dc motor drive,” in Industrial Electronics and Applications (ICIEA), 2011 6th IEEE Conference on, june 2011, pp. 2723 –2728. [8] H. Ransara and U. Madawala, “A low cost drive for three-phase operation of brushless dc motors,” in IECON 2011 - 37th Annual Conference on IEEE Industrial Electronics Society, nov. 2011, pp. 1692 –1697. [9] C. Luongo, P. Masson, T. Nam, D. Mavris, H. Kim, G. Brown, M. Waters, and D. Hall, “Next generation more-electric aircraft: A potential application for hts superconductors,” Applied Superconductivity, IEEE Transactions on, vol. 19, no. 3, pp. 1055 –1068, june 2009.

1908