Anda di halaman 1dari 8

Closed Loop Control of Excitation Parameters for High Speed

Switched-Reluctance Generators
Yilmaz Sozer David A. Torrey

Advanced Energy Conversion, LLC


31 Ontario Street
Cohoes, NY 12047 USA

Abstract energy application is characterized by low speed, high torque


operation. Standard references on the switched-reluctance ma-
This paper presents a new approach to the automatic control chine (SRM) are given in [9, 10]; control of the SRG is dis-
of excitation parameters for the switched-reluctance generator cussed in [11].
(SRG) where the SRG system operates at sufficiently high speed The SRM produces torque through excitation that is synchro-
that it operates in the single pulse mode. The turn-on and turn- nized to rotor position. The excitation is generally described
off angles are the two parameters through which we can con- by three excitation parameters: the turn-on angle on , the turn-
trol the electric power generation. The objective of the work off angle o , and the reference current Iref . With operation
is to develop an easily implementable control algorithm that above based speed, current regulation is not possible and on
automatically maintains the most efficient excitation angles in and o are the only two control parameters that prescribe the
producing the required amount of electric power. The work is current waveform. A control algorithm would typically use the
focused on finding the most efficient excitation angles and char- same excitation parameters for each phase, implemented with
acterizing them for easy implemention under closed loop con- the spatial shift consistent with the symmetrically displaced
trol. Through modeling of an experimental SRG and extensive phase structure. Control of the excitation angles results in ei-
simulation, it can be seen that the optimal-efficiency turn-off ther positive net torque for motoring, or negative net torque for
angles can be characterized as a function of power and speed generating.
level. Within the closed-loop power controller, the optimal-
efficiency turn-off angle is determined from an analytic curve Control of the SRM is more complicated for generator operation
fit. The turn-on angle is then used as the degree of freedom nec- than it is for motor operation. Operating as a motor, the turn-
essary to regulate the power produced by the SRG. Given that on angle can be used to directly control the peak phase current,
the turn-off angle is associated with optimal-efficiency at each thereby partitioning the responsibilities of the turn-on angle and
speed and power point, overall operation is achieved at optimal- conduction angle (cond = o on ). Operating as generator,
efficiency. The SRG, inverter and control system are modeled the turn-on and turn-off angles control the peak phase current
in Simulink to demonstrate the operation of the system when jointly and severally. However, we have no control over where
implemented within a voltage regulation system. The control the peak phase current occurs once the SRG has entered single
technique is then applied to an experimental SRG system. Ex- pulse operation above base speed. This leaves us with many
perimental operation documents that the technique provides for combinations of turn-on angle and conduction angle that will
efficient operation of the SRG system through tuning the con- yield the same output power. An important issue is how to best
troller at only four operating points. choose the excitation parameters. This is more challenging for
the SRG than the SRM because the peak phase current for the
SRG occurs while both controllable switches are off.
1 Introduction Efficient operation of the SRM, or any motor drive, is al-
ways of importance. Inefficiency leads to larger size, increased
weight, and increased energy consumption. In order to maxi-
The switched-reluctance generator (SRG) is under development
mize switched-reluctance generator (SRG) efficiency, we seek
for variable speed applications where the inherent characteris-
to maximize the ratio of the average output power to given input
tics of the SRG make commercial sense. To date, these ap-
average mechanical power, Pout /Pmech. This ratio captures our
plications include sourcing aerospace power systems [1, 2, 3],
intended goal of providing the required electrical output with
starter/alternators for hybrid vehicles [4, 5, 6], and wind turbine
the minimum electromechanical input.
applications [7, 8]. The aerospace and automotive applications
are generally characterized by high speed operation. The wind This paper presents a closed loop power control algorithm for

0-7803-7768-0/03/$17.00 (C) 2003 IEEE 75


3000 3000

2500 2500

2000 2000

Dc Power (W)
DC Power (W)

1500 1500

1000 1000

500 500

0 0

500 500
140 120 100 80 60 40 20 0 100 50 0 50 100 150 200
Turn on Angle (Elec Degree) Turn Off Angle (Elec. Degree)

Figure 2: Dc power generation versus turn-on angles at Figure 3: Dc power generation at 3000 rpm for all possi-
3000 rpm for all possible excitation angles. ble combinations of excitation angles, shown as
a function of turn-off angle.

SRG with most efficient excitation angles. The method does


not need machine modelling or extensive simulations. The al- by the phase current and its relative position with respect to in-
gorithm is easy to implement and does not need look up ta- ductance profile. In order to produce a map of power production
bles for excitation parameters. Implementation of the algorithm for a given speed, turn-on and turn-off angles, the SRG is mod-
is demonstrated in simulation and experiment for a four phase elled using finite element magnetization data and the generator
SRG when implemented as part of a voltage regulation system. system is simulated for every 1000 rpm between 3000 rpm and
10,000 rpm with turn-on and turn-off angles taking all combi-
The algorithm is given in Sec. 2. The operation of the algorithm nations of on and o in increments of 2 electrical degrees.
is verified through simulation in Sec. 3. Section 4 provides op- Figures 2, 3, and 4 show dc power generation versus excitation
erational verification on a 1.8 kW experimental SRG drive. angles at 3000 rpm. Among all simulated data, the optimum
turn-on and turn-off angles are selected for a given speed and
power level. The criteria is to find the excitation angles that
2 The Algorithm produce the required power at a given speed with minimum me-
chanical power. Figures 5 and 6 show the optimum excitation
angles for a given speed and power level.
The objectives of the algorithm are best explained through con-
sideration of the linear inductance profile for the SRM shown in It would require an extensive look up table to experimentally
Fig. 1. The minimum inductance region is defined by the angu- implement the optimum excitation angles, not to mention the
lar interval over which the rotor poles do not overlap the stator time required to experimentally determine the optimum values
poles. The maximum inductance region is defined by the an- of on and o . Also due to the discrepancy between the actual
gular interval over which there is complete overlap between the system and the motor model and changes in system parameters
stator and rotor poles. The regions of increasing and decreas- during the operation of the system, it is very difficult to achieve
ing inductance correspond to varying overlap between the stator power control under open loop conditions prescribed by a look
and rotor poles. up table.
For operation of the SRM as a motor, phase current must be As we have seen, the optimum turn-off angles are a function of
present in the phase winding as the inductance is increasing in operating speed and average output power. The optimum turn-
the direction of rotation. For operation of the SRM as a gener- off angles can be represented as
ator, phase current must be present in the phase winding as the
inductance is decreasing in the direction of rotation. The polar- o = k1 p + k2 + k3 p + k4 , (1)
ity of current is immaterial, so we assume that the phase current
where is rotor speed, p is the output power and k1 , k2 , k3 and
is always positive.
k4 are curve fit parameters. For our simulation work the curve
As we operate the generator above base speed, the magnitude fit parameters are based on a least squares fit to the collection
and shape of the phase current directly depends on the turn-on of optimal turn-off angles over all operating points above base
and turn-off angles. The amount of power generated is dictated speed. For our experimental work the curve fit parameters are

76






rotation
0 g 180 m 360

stator
rotor

Figure 1: The linear inductance profile of the SRM show-


ing g and m .

3000 120

2500 110
Turn off (Elec. Degree)

2000 100
Dc Power (W)

1500 90

1000 80

500 70

0 60

500 50
160 2000

140 0 1500 10000


20 9000
120 1000 8000
40
60 7000
100 6000
80 500 5000
80 100 4000
120 0 3000
60 140 Power (W)
Conduction Angle (Elec. Degree) Speed (rpm)
Turn on Angle (Elec. Degree)

Figure 4: Dc power generation versus turn-on and con- Figure 5: Turn-off angles versus speed and dc power for
duction angles at 3000 rpm. an optimized excitation angles.

77
120 100

100
90
Turn On (Elec. Degree)

Turn Off (Elec. Degree)


80
80
60
70
40

60
20

0 50
10000 10000

8000 2000 8000 2000


1500 1500
6000 6000
1000 1000
4000 4000
500 500
2000 0 2000 0
Speed (rpm) Speed (rpm)
Dc Power (W) Dc Power (W)

Figure 6: Turn-on angles versus speed and Dc power for Figure 8: Turn-off angles versus speed and dc power with
an optimized excitation angles. a fit to Eq. 1 using only four points from the
optimized data.

120
95

90

85 110
Turn Off (Elec. Degree)

80

75
100
Turn Off Angle (Elec. Degree)

70

65
90
60

55

50 80
10000

8000 2000
1500
6000 70
1000
4000
500
2000 0
Speed (rpm) 60
Dc Power (W)

3000 rpm 4000 rpm 5000 rpm 6000 rpm 7000 rpm 8000 rpm 9000 rpm 10000 rpm
Figure 7: Turn-off angles versus speed and dc power with 50
0 50 100 150
a least squares fit to Eq. 1 using all optimized
data. Figure 9: A comparison of the curve fits for turn-off an-
gles. Stars show all of the optimized turn-off
angles. Solid lines show the with a least squares
based on the optimized data for four operating points represent- fit to Eq. 1 using all of the turn-off angles. Dia-
ing all combinations of low speed, high speed, low power and monds show the curve fit using only four points
high power. Figures 7, 8, 9 show the curve fit of the turn-off an- from the optimized data.
gles using all data and data from only four operating points. As
seen from these figures the curve fit using only four data gives
results that are close to the actual optimized turn-off angles.
As we explained earlier, only the turn-on and turn-off angles trol of output power is maintained while enforcing optimal ef-
have control over power generation for operation above base ficiency. This is based on the premise that while there are mul-
speed. From the above studies we can get the efficiency-optimal tiple (on , o ) combinations that will provide the same output
turn-off angles easily. The closed loop power control can then power at a particular operating speed, only one pair will yield
provide the turn-on angle. Since the turn-off angle is optimum the highest efficiency. Figure 10 shows the closed loop power
for a given speed and power, the turn-on angle being output control algorithm. Adjustment of on is constrained to prevent
by the power controller is then guaranteed to be efficiency- continuous conduction of the phase currents; on represents the
optimal as well. Approached in this manner, closed-loop con- limit on on .

78
idc

vdc

p



p
r+

on
Table 1: The specifications for the experimental SRG.

on Quantity Value Units
Rated Power 1800 W
cond
Base Speed 2000 rpm
Maximum Speed 10,000 rpm
Curve + o Dc Voltage 14.8 V
Fit r Number of Rotor Poles 12

(Eq. 1) Number of Stator Poles 16
Number of Phases 4
Aligned Phase Inductance 0.2290 mH
Unaligned Phase Inductance 0.0175 mH
Figure 10: The algorithm used to automatically adjust the g 132
(electrical)
excitation angles.
m 228 (electrical)

3 Verification through Simulation


The algorithms motivated in Sec. 2 were implemented in simu-
lation to confirm proper operation before being experimentally
implemented on the physical system. The SRG to which the
simulation is applied is a 16/12 four-phase SRM designed for a
1.8 kW 12 V automotive application; Table 3 gives the parame-
ters of the SRG used in this work. The SRG magnetics are mod-
eled analytically based on data collected through finite element
analysis. Figure 11 shows the result of the proposed power con-
Speed (rpm)

4000

trol technique for 1600 W reference power and varying speed 2000

profile. Figure 12 shows the simulation results of the imple- 0


0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45
mented control technique where speed is ramped to 5000 rpm
Turn Off (Elec. Deg.)

70

in 0.15 s and kept constant. The power reference is changed 60

from 1600 W to 300 W at 0.25 s. These figures show the con- 50

trol technique provides the commanded power by adjusting the 40


0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45
excitation angles. To the extent that o corresponds to that re-
Turn On (Elec. Deg.)

50

quired for peak efficiency based on the control algorithm, the 60


70
required power is produced with the highest efficiency possible. 80
90
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45
2000
Dc Power (W)

4 Experimental Results 1000

0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45
Time (s)
The performance of the controller is experimentally verified
through a 16/12 four-phase SRG designed for a 1.8 kW 12 V
automotive application. The control algorithm that we have de- Figure 11: Simulation results of the implemented control
veloped is implemented using an Analog Devices ADMC401 technique with 1600 W reference power with a
digital signal processor. The SRG is coupled to an induction varying speed profile.
motor, which acts as constant speed mechanical power source
using an adjustable speed drive. A resolver with resolver to dig-
ital converter circuitry provides 12 bit position information to
the DSP. The 12 V battery and load resistors are connected in
parallel to the inverter. Figure 13 shows the block diagram of

79
Phase A
Phase B

Phase C
1
0 1
0 1
0 1
0 1
0 1
0 1
0 IM
0
1 0
1 0
1 0
1 0
1 0
1 0
1

+ Vdc
1
0 11
00 Phase D
- 11
00 0
1 1
0 11
00 00
11 1
0 SRG
Load Battery 0
1

1
0 1
0 1
0 1
0 1
0 1
0 1
0 Current
Sensing Resolver
Voltage
Sensing S1 S2 S3 S4 S5 S6 S7 S8
ia
ib
Analog Circuitry ic
id

S1 S2 S3 S4 S5 S6 S7 S8

Vdc
Digital Signal Processor idc

Figure 13: The block diagram of the experimental setup.

55

50
Turnoff Angle (Elec. Deg.)
Speed (rpm)

4000
45
2000

40
0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45
Turn Off (Elec. Deg.)

70
35
60

50 30
10000
40
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 8000 2000
Turn On (Elec. Deg.)

40 1500
6000
1000
60 4000
500
2000 0
80 Speed (rpm)
Dc Power (W)

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45


2000
Figure 14: Experimentally found input shaft torque ver-
Dc Power (W)

1000 sus turn-off angle at 3000 and 6000 rpm with


0
low and high power cases.
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45
Time (s)

Figure 12: Simulation results of the implemented con- the experimental setup for the tests.
trol technique speed is ramped to 5000 rpm in
0.15 s and kept constant. The power reference We experimentally found the optimum turn-off angles for four
is changed from 1600 W to 300 W at 0.25 s. operating points representing all combinations of low and high
speed with low and high power. Figure 15 shows the input shaft
torque versus turn-off angles at 3000 and 6000 rpm with low
and high power levels. The turn-off angle providing the desired
power with the minimum torque is the most efficient one.
Having the most efficient excitation angles for four operating
points we determine the turn-off angles for any given point us-

80
3000 rpm, 650 W 3000 rpm, 1800 W
3.4 8.4

3.3
Shaft Torque (Nm)

Shaft Torque (Nm)


8.2

3.2
8 400

Phase Current (A)


3.1
200
7.8
3
0
2.9 7.6
0 20 40 60 80 20 30 40 50 60 70
Turnoff Angle (Elec. Deg.) Turnoff Angle (Elec. Deg.) 200
0 1 2 3 4 5 6 7 8 9 10
6000 rpm, 650 W 6000 rpm, 1500 W
3000
1.52 3.45

Output Power (W)


2000
1.5 3.4
Shaft Torque (Nm)

Shaft Torque (Nm)


1000
1.48
3.35
0
1.46
3.3 1000
1.44 0 1 2 3 4 5 6 7 8 9 10

1.42 3.25 16

Dc Bus Voltage (V)


1.4 15
3.2
20 30 40 50 60 70 45 50 55 60 65 70
Turnoff Angle (Elec. Deg.) Turnoff Angle (Elec. Deg.) 14

13

Figure 15: The variation in shaft torque with turn-off an- 12


0 1 2 3 4 5 6 7 8 9 10
gle determined experimentally. Time (s)

Figure 17: The response of the experimental system


where the load is changed from 300 W to
Table 2: The parameters for the controllers of Figs. 10 and 16.
1600 W.
Quantity Value Units

Kp (Power) 0.033 /W (electrical)

Ki (Power) 1.899 /J (electrical)

cond 178 (electrical)

k1 7.41e-7 s/W (electrical)

k2 5.55e-4 s (electrical)

k3 4.44e-3 /W (electrical)

k4 31.33 (electrical)
Generator Speed (rpm)

5000

Kp (Voltage) 140.65 W/V 4000

Ki (Voltage) 1620.80 W/V s 3000

2000

1000
0 1 2 3 4 5 6 7 8

2000
Output Power (W)

ing Eq. 1. Figure 14 shows the dependence of shaft torque 1500

on turn-off angle using experimental data from the four oper- 1000

ating points. The turn-off angles of interest are those that mini- 500

mize shaft torque while supporting the required output power at 0


0 1 2 3 4 5 6 7 8
speed. 15
Dc Bus Voltage (V)

Voltage control is implemented as an outer control loop. The 14.5


voltage controller commands the reference power necessary to
14
drive the output voltage to the required level. In order not to
harm mechanical components and to control the charging rate 13.5
0 1 2 3 4 5 6 7 8
of the battery, a rate limiter is implemented. The rate limiter Time (s)

interacts with the PI controller to keep the integrator out of sat-


uration. Table 2 provides the parameters used in the voltage and Figure 18: The response of the experimental system
power controllers. Figure 16 shows the block diagram of the where the speed is changed from 2000 rpm to
voltage controller applied experimentally. Figure 17 shows the 4000 rpm.
response of the experimental system where the load is changed
from 300 W to 1600 W. The rate of change of power is limited
to 600 W/s. Figure 18 shows the response of the system where
the speed is changed from 2000 rpm to 4000 rpm.

81

vdc + Power p
Rate Limiter






vdc

Figure 16: The algorithm used to control the dc bus volt-


age.

5 Summary ator, Proc. of the IEEE Applied Power Electronics Conf.,


pp. 362-368, 2000.
The behavior of the switched-reluctance generator efficiency [6] M. Besbes, M. Gabsi, E. Hoang, M. Lecrivain, B. Gri-
is characterized in terms of operating speed, power production oni and C. Plasse, SRM design for starter-alternator sys-
and excitation angles. The most efficient turn-off angles are a tem, Proc. of the International Conference on Electric
function of power level and speed, so they can be represented Machines, pp. 1931-1935, 2000.
through a curve fitting function. The turn-on angles are the out-
put of the closed loop power control. Having the most efficient [7] D. A. Torrey, Variable-reluctance generators in wind-
turn-off angle, we can say that turn-on angle which provides the energy systems, Proc. of the IEEE Power Electronics
required power level for that operating condition is the most ef- Specialists Conf., pp. 561-567, 1993.
ficient one. The overall system and control is simulated with the
SRG model based on finite element data. The control technique [8] R. Cardenas, W. F. Ray and G. M. Asher, Switched re-
is implemented on an experimental system as part of an overall luctance generators for wind energy applications, Proc.
voltage controller. The new technique provides easy implemen- of the IEEE Power Electronics Specialists Conf., pp. 559-
tation of the control technique. The excitation angles are easily 564, 1995.
optimized by determining the efficiency-optimal excitation pa- [9] P. J. Lawrenson, et al., Variable-speed switched reluc-
rameters at only four operating points. The results of these tests tance motors, IEE Proc., Vol. 127, pt. B, no. 4, pp. 253-
show that the new control technique provides an efficient SRG 265, 1980.
controller that is easy to implement.
[10] T. J. E. Miller, Switched reluctance motors and their con-
trol, Oxford University Press, 1993.
References [11] D. A. Torrey, Switched-reluctance generators and their
control, IEEE Trans. on Industrial Electronics, Vol. 49,
[1] S. R. MacMinn and J. W. Sember, Control of a switched- pp. 3-14, 2002.
reluctance aircraft starter-generator over a very wide speed
range, Proc. of the Intersociety Energy Conversion Engi- [12] D. A. Torrey and J. H. Lang, Modelling a nonlinear
neering Conf., pp. 631-638, 1989. variable-reluctance motor drive, IEE Proc., Vol. 137, pt.
B, pp. 315-326, 1990.
[2] A. Radun, Generating with the switched-reluctance mo-
tor, Proc. of the IEEE Applied Power Electronics Conf.,
pp. 41-47, 1994.

[3] D. E. Cameron and J. H. Lang, The control of high-speed


variable-reluctance generators in electric power systems,
IEEE Trans. on Industry Applications, Vol. 29, pp. 1106-
1109, 1993.

[4] J. M. Kokernak, D. A. Torrey and M. Kaplan, A switched


reluctance starter/alternator for hybrid electric vehicles,
Power Electronics Proc. (PCIM Conference), pp. 74-80,
1999.

[5] E. Mese, Y. Sozer, J. M. Kokernak and D. A. Torrey, Op-


timal excitation of a high speed switched reluctance gener-

82

Anda mungkin juga menyukai