Deadbeat Flux Level Control Of Direct Field-oriented High Horse Power Induction Servo Motor Using Adaptive Rotor Flux Observer

IEEE TRANSACTIONSON INDUSTRY APPLICATIONS,VOL. 30, NO. 4,JULY / AUGUST 1994

954

Deadbeat Flux Level Control of Direct-Field-Oriented High-Horsepower Induction Servo Motor Using Adaptive Rotor Flux Observer Lipei Huang, Yasuki Tadokoro, and Kouki Matsuse, Senior Member, IEEE, Abstract-A method for the deadbeat flux level control of direct-field-oriented high-power induction servo motor drives has been developed that employs an adaptive rotor flux observer. The observer is a full-order type and is used not only in the direct-field-oriented controller, but also to determine the stator and rotor resistances of the servo motor. The observer reduces the sensitivity of a deadbeat controller to fluctuations in the motor parameters. The main advantage of this method is that it improves the efficiency of an induction servo motor without sacrificing dynamic performance. This paper describes the method as well as the fundamental characteristics of the system derived from experimental and simulation results.

I. INTRODUC~ON

T

HE indirect-field-oriented control of an induction servo motor is widely used to obtain good dynamic torque response. The system generally keeps the rotor flux level constant, but it has the disadvantage of being very sensitive to fluctuations in the motor parameters. In order to overcome this drawback, a direct-field-oriented control method employing an adaptive rotor flux observer has recently been proposed [ll, 121. On the other hand, it has been reported that changing the rotor flux level in accordance with the load improves the steady-state efficiency of an induction motor [31-[5]. But those papers do not discuss the transient speed and torque response. To obtain higher steady-state efficiency without degradation of the dynamic response, a method of rapidly controlling the rotor flux level by means of a deadbeat response has been proposed [6]-[8]. Based on these considerations, a deadbeat flux level control system applicable to direct-field-oriented induction servo motor drives has been developed that employs Paper IPCSD 94-28, amroved by the Industrial Drives Committee of the iEEE Industry Ap&cations Society for presentation at the 1993 IEEE Industry Applications Society Annual Meeting, xxx, m, Date. Manuscript released for publication April 1, 1994. L. Huang is with the Department ofElectrica1 Engineering, Tsinghua University, Beijing 100084,-China. Y. Tadokoro was with the Department of Electrical Engineering, Meiji University, Higashimita, Tama-ku, Kawasaki 214 Japan. He is now with the Ebara Comration. Power Electronics DeDartment, 2-1, Honfujisawa 4-chome, Fuj’isawa-shi 251, Japan. K. Matsuse is with the Department of Electrical Engineering, Meiji University, Higashimita, Tama-ku, Kawasaki 214 Japan. IEEE Log Number 9402640.

an adaptive rotor flux observer. The observer is a full-order type, and has the additional function of determining the stator and rotor resistances of the motor on the basis of adaptive control theory. This paper discusses the operational efficiency of a servo motor driven by the proposed system in both steady and transient states, and also the dynamic performance. The load cycle dependency is also considered, as is appropriate for high-horsepower induction servo motors. This is important because increasing the rotor flux level at moments of high torque demand requires extra current, and consequently, extra loss. The principles and fundamental characteristics of the system are presented and discussed on the basis of simulation and experimental results. 11. DEADBEAT FLUX LEVELCONTROL SYSTEM A. Deadbeat Control of the Rotor F l u Level TakingMagnetic Saturation into Account One way to improve the steady-state efficiency of a motor is to change the magnitude of the rotor flux in accordance with the topic required. To achieve this, the approach taken in this study is to employ a deadbeat response to regulate the rotor flux level. This should yield a design for a digital controller that produces a flux level response with zero steady-state error and has a minimum finite settling time in response to a unit step input. In a rotating frame of reference, the rotor flux, can be expressed in terms of the flux current, i,, by (1) taking magnetic saturation into account:

+,,

where & is the rotor flux, i , is the flux current, M is the mutual inductance, L , is the’rotor self-inductance, and R , is the rotor resistance. In practice, the effects of magnetic saturation can have a significant impact on rotor flux level control because both the rotor time ‘Onstant, ‘r7 and the induetanCe, M , Vary ill the Operating region O f the motor. Tables I and I i list the rafings ofthe-two induction servo motors tested, and Fig. 1 shows the characteristics of servo motor I. Fig. l(a> shows the relationship between

0094-9994/94$04.00 0 1994 IEEE

955

HUANG et al.: DEADBEAT FLUX LEVEL CONTROL

TABLE I RATINGSOF SERVO MOTOR 1

2.2 kW

10.5 A 120 kg-cm

170 V

2000 rpm

4 pole

TABLE I1 RATINGSOF SERVOMOTOR11

A

37 kW 4 pole

184.0 A 2300 kg-cm

140 V 50 Hz

0

2

6

4

where * indicates reference value, indicates estimated value, T is the settling time, ~ , [ i T lis the rotor time constant (7, = L , / R , ) at sampling time iT, M[iTI is the mutual inductance at sampling time i T , +,[iT] is the flux level at sampling t i p e iT, and i equals n - 2, n - 1, n. The flux level, +,, is estimated by the adaptive rotor flux observer, which also determines the stator and rotor resistances on the basis of adaptive control theory. Fig. 2 illustrates how deadbeat flux level control works. In a conventional linear system, Fig. 2(a), the flux settles by an amount A+, during the settling time T . But in the proposed method, Fig. 2(b), the flux settles in a series of steps by an amount A+; during the subsettling time T'. This is because the flux current is limited to its maximum value if the change in the rotor flux is too large. In this study, T' was set to 7.5 msec based on the CPU speed. In Fig. 2(b), the dotted lines A and B show the effects of magnetic saturation.

8

10

12

im (A)

(a) Rotor Flux (e,) vs. Flux Current (i,,,) 0.048

B. Adaptive Rotor Flux Observer

0.046

g 0.044 2 0.042 0.040

0.038

I).

1

0.25

0

0.30

.

0.35

1

0.40

(Wb)

Eqs. (3) and (4) are the state and output equations, respectively, for a stationary frame of reference. Eq. ( 5 ) describes the full-order state observer, which simultaneously estimates the stator current and the rotor flux. Eq. (6) is the observer gain matrix, which is calculated in such a way that the observer poles are proportional to those of the induction motor (with a constant of proportionality, k > 0). An induction motor itself is stable, so the adaptive observer is also stable under normal operation:

(b) Mutual Inductance ( M ) vs.Rotor Flux (e,) Fig. 1. Characteristics of servo motor I.

rotor flux and flux current, and Fig. l(b) shows the relationship between mutual inductance and rotor flux. In the latter, it should be noted that the mutual inductance is constant when the rotor flux is less than 0.25 Wb. This figure demonstrates that the mutual inductance changes in the region of magnetic saturation, and so its effects must be taken into consideration when controlling the flux [9]. Accordingly, the reference flux current, i:, can be given as follows [8]: 1 i:[nT] = 1 - exp ( - T/T,[n T ]

. +,*[nTl ~

[

-

M[nTl

&[nTIM [ ( n - 1)Tl

1

exp ( - T / T , [ ( ~ 1)TI) 1 - exp ( - T / T , [ ( ~ 1)TI)

i = CX,

where

I=[;

;I

J = i0l -11

R, stator resistance L , stator self-inductance a leakage coefficient ( a = 1 - M 2 / ( L , L , ) ) w, motor angular velocity

(4)

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 30, NO. 4, JULY / AUGUST 1994

956

(a) d-axis Equivalent Circuit n

l

t

(a) Linear System

(b) Proposed System

Fig. 2. Deadbeat flux level control.

and d -2 dt

=

i- BU,

+ ~ ( i '-, i s ) ,

(b) q-axis Equivalent Circuit Fig. 3. Steady-state equivalent circuits taking core loss into account.

(6)

g,

=

(k

g3

=

-(k2

- l)(UrI,

+ ur,,)

- l)(UrlI - U r 3 )

g,

=

(k -

m+

1)Ui2,

+ c ( k - l)(Ur1l + UrZ2)

- Or @qr

111. LOSSES I N STEADY AND TRANSIENT STATES For steady-state operation, the d- and q-axis equivalent loss into circuits of an induction motor that take account can be drawn as in Figs. 3(a) and 3(b), respectively. These circuits yield the following equations for the stator copper loss, the rotor copper loss, and the stator core loss in the steady state:

w,,= R S & ~2

=

1- i & ) ,

(9)

Rr(i2r + i i r ) ,

(10)

U: = R,,,((ids+ i d r l 2+ ( i q r+ i q r ) 2 ) , where i,

=

[idr i,,lT is the rotor current.

(11)

For transient-state operation, the circuits are shown in Figs. 4 a ) and 403). Here, the core loss is accounted for by the eddy current loops [lo]. These circuits yield the following equations for the and the core loss [111:

9

where i,

=

[ i d , i,,lT

eddy current

(12)

~

957

HUANG et al.: DEADBEAT FLUX LEVEL CONTROL

h

l

o

o

n

$r=0.250Wb Constant p o . 3 0 0 Wb Constant p=0.350 Wb Constant .... +r=0.400 Wb Constant .......

W

0

20

I

I

I

I

40

60

80

120

100

Load Torque (kg-cm) (a) Efficiency

- U Deadbeat Flux Level Control p=0.250 Wb Constant

Fig. 5. Block diagram of proposed control system for application to direct-field-oriented induction motor drives. (CR: current regulated.)

0

80

40

120

Load Torque (kg-cm) Fig. 7. Motor efficiency and total loss for servo motor I (2.2 kW, ac).

,

100

I

Fig. 6. System configuration.

Zll

=

-

( R , + PL:)Z

Deadbeat Flux Level Control p=0.200 wb Constant ...... p=0.250Wb Constant +1=0.300 Wb Constant - - - p S . 3 5 0 Wb Constant .. .......

PM'Z 2 2 1 = 2 2 3 = PM'Z - w,M'J 2 2 2 = ( R , + PL:)Z - W,Z, J Z3, = ( R e + PM')Z R , core loss resistance d p = dt and 212 =

Z,3

= 2 3 1 = Z32 =

0

500

lo00

1500

2000

Load Torque (kg-cm)

,

3000 (13)

where

5

2000-

A

3

1000-

0

where To is the speed command cycle (i.e., load cycle).

I

-m Deadbeat Flux Level Conwol -*- @=0.350 Wb Constant

0

The total motor loss for transient-state operation is given by (14). Finally, (15) gives the average motor loss, which includes the loss during both steady- and transientstate operation, during one speed command cycle, or in other words, one load cycle:

(a) Efficiency

I

I

I

500

1000

1500

2000

Load Torque (kg-cm) Fig. 8. Motor efficiency and total loss for servo motor I1 (37 kW,ac).

IV. EXPERIMENTAL AND SIMULATION RESULTS A. System Configuration The configuration of the proposed deadbeat rotor flux level control system is shown in Fig. 5. This system is used for the direct-field-oriented control of an induction servo motor 111, [21. The main components are an adaptive rotor flux observer; a deadbeat controller; a numerical function

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 30, NO. 4, JULY / AUGUST 1994

958

500rpn;/div T 10A/div

0'

0 '

:'!

:

:

(a) Simulation Results

:

:

:

50 m S e C / d ~

' 1V

500r pm/d i v T

r-----

0.

0 . 1 OWb/d i v

---

T

qr -ir

-.

'

-

__

t

%

50msec/clv

-.-idr -..-iqr 0

0 I

(a') Simularion Results

(a') Simulation Results

0

0 (b) Expenmental Results

( b ) Expenmental Results

Fig. 10. Step-up speed response-stator current (0

Fig. 9. Step-up speed response-motor speed and flux level (0 rpm).

+

+

I000 rpm).

1000

+,:

operator, F , that sets the appropriate flux command, in response to the torque command, T: ; a direct-fieldorientation controller; and a current-regulated voltage source inverter. Fig. 6 shows the experimental setup for the digital signal processor and CPU.

___

200rpm/div

r*

-U ,

B. Direct-Field-Oriented Controller Using an Adaptive Flux Obsemer In the vector rotation block in Fig. 5, the stator current commands, i2s and i:s, for a stationary frame of reference are calculated from the estimated rotor flux as follows [l]:

e", sin e^ + i: cos e",

izs = i: cos

i:s

=

i:

o - iT

sin

&/& 4,= d

where cos e" = O d r / J r , sin e" = and iT is the torque current command.

0 . 1 OW s / d

1v

_--

q * -Fr

---I -..-$ dr

4'

(16)

(17)

10A/div

m ,

i,

-

/ -

C. Steady-State EfJiciency and Loss

Figs. 7(a) and 7(b) show the relationship between load torque and motor efficiency, and that between load torque and total loss, respectively, of servo motor I. Figs. 8(a) and 8(b) show the same relationships for the high-horsepower servo motor 11. The dashed and dotted curves are for various values of the rotor flux level, and the heavy black curves show the characteristics of the motors when driven by the proposed deadbeat control system. It is clear that

I

104/div

0

Fig. 11. Step-up speed response-experimental (proposed method).

HUANG er al.: DEADBEAT FLUX LEVEL CONTROL

959

-Deadbeat Flux Level Control @=0.400[Wb] Constant

200rpm/d i v

--____

(a) Speed

10A/div

-ids

0 1

(b) @i-=O.250-0.400[Wb] Deadbeat Flux Level Control

IOA/div

T

~

ids

0 ( c ) @CO 400 Constant Rotor Flux

1O A / d 1 v

10A/div

__

-Ids

Ids I

0

(d) @r=0.250Constant Rotor Flux

Fig. 14. Comparison of proposed method and two cases of constant rotor flux method. (The data are from Figs. 11-13.)

Fig. 12. Step-up speed response-experimental

(constant

4,

=

0.400

=

0.250

wb).

___ Ur* _100rpm/div T

r--------

-U , --

1 OA/d i v

T

.* -1,

I

Fig. 13. Step-up speed response-experimental wb).

(constant

Fig. 15. Ramp speed response-experimental (500

-, -500

vm).

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 30, NO. 4, JULY / AUGUST 1994

960

o p p

100rpm/div T

5 0T 0 r pm/d i v

_ _ _U:

OOmsec/div

O.lOWb/div

I

1

.*

-[in

10A/div

T

---10A/div

.*

5A/d i v

If

500msec/div

-Ids

0

0

Fig. 16. Load torque response-experimental.

5 0 0 m s e c Id i

Fig. 18. Forward-reverse speed response-simulated 0.400 Wb).

500rpm/div

--_

500rpm/div

U,'

/OOmsec/div O.IOWb/div

5A/d i v

1 OOW/d iv

i

0

---$r

___

U

(constant

4r =

,*

,

-qr

/OOmsec/di\j I

O.lOWb/div

---$I

4Jr

~

.* 11

-Loss

500msec/div

Fig. 17. Forward-reverse speed response-simulated (proposed method).

'l O O W /d i v

I

0

~

Loss

5 0 0 m s e c/ d 1v

Fig. 19. Forward-reverse speed response-simulated 0.250 Wb).

(constant

c$~ =

96 1

HUANG el al.: DEADBEAT FLUX LEVEL CONTROL

'

'

(a) Condition [I]

(b) Condition [2]

4'(rpm)

or'(rpm)

I

(c) Condition [3]

(d) Condition [4]

I (e) Condition [ 5 ]

300(rpm) Constant Command (g) Condition [ 7 ]

(0 Condition [6]

Fig. 20. Speed command conditions (i.e., load cycles).

TABLE I11 COMPARISON OF LOSSES

our method significantly improves the efficiency while reducing the total loss.

0.400 Wb beadbeat Constant Control Loss 1 Loss 1 Loss 2 To (W) (w) Loss2 (s)

Constant Loss 3 Loss 1

2.0

80.398

Condition

D.Transient Response Figs. 9 and 10 show the experimental and simulation results for the step-up speed response of servo motor I when using the proposed control method. The speed command was a step function with values in the range of 0-1000 rpm, and there was no load. Figs. 11-13 show the experimentally obtained speed, rotor flux, command current, and stator current for a step-up speed change operation on servo motor I. Fig. 11 is for our proposed method, and Figs. 12 and 13 are for the conventional constant flux control method. The speed command was a step function with values ranging from -500 to 500 rpm, and there was no load. In each case, the command torque current, if*, was set to the upper transient-state limit of the motor. Fig. 14 combines these results for easier comparison, with Fig. 14(a) showing the speed, and Figs. 14(b)-14(d) showing the currents. Figs. 7, 8, and 14 clearly demonstrate that driving an induction servo motor with the proposed system improves the steady-state efficiency without sacrificing the dynamic performance. Fig. 15 shows the experimentally obtained forwardreverse speed response of servo motor I. The speed command was a ramp function with values ranging from 500 to - 500 rpm, and there was no load. Fig. 16 shows the experimental results on the load torque response of servo motor I. The load torque changed in steps from 0 to 60 kg-cm.

E. Discussion of Steady and Transient Losses Figs. 17-19 show the simulation results on the forward-reverse speed change response of servo motor I.

0.250 Wb

(W)

Loss 3

I

I

[7]

65.744

89.055

0.738

0.818

Fig. 17 is for our proposed method, and Figs. 18 and 19 are for the constant flux method with values of +r of 0.400 Wb and 0.250 Wb, respectively. These results enable us to compare the total loss of a motor when driven by the different methods. The loss was calculated using (91-0 11, (13) and (14) for both the steady and transient states. Fig. 20 shows the various speed command conditions, that is, load cycles, for servo motor I that were put into (15) to calculate the results in Table 111, which compares the average loss obtained for our proposed method and for the constant flux method with the two values of the flux mentioned in the previous paragraph. The table reveals that our proposed method yields a significantly lower loss except for condition 4. In this case, the constant flux method with a of 0.400 Wb gives better results.

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 30, NO. 4, JULY / AUGUST 1994

962

V. CONCLUSION A deadbeat flux level control method for direct-fieldoriented induction servo motor drives that employs an adaptive rotor flux observer has been developed. The fundamental characteristics of the system were investigated through experiments and simulations. It was found that the application of this system to high-horsepower induction servo motors improves the efficiency in both the steady and transient states without sacrificing dynamic performance. ACKNOWLEDGMENT The authors would like to thank M. Tsukakoshi of Meiji University, N. Ishii of Tokyo Electric Power C o . , Ltd. and K. Miyamoto of JR Tokai Co., Ltd. for their support of the experimental work. They also wish to thank Dr. H. Kubota of Meiji University and K. Kikuchi of Sanyo Denki Co., Ltd. for useful discussions. REFERENCES Ill [21 I31

I41

I51

[111

H. Kubota, K. Matsuse, and T. Nakano, “New adaptive flux observer of induction motor for wide speed range motor drives,” in IEEE, IECON’90, Proc., vol. 11, pp. 921-926, 1990. H. Kubota and K. Matsuse, “Adaptive flux observer of induction motor and its stability,” Trans. IEE Jpn., vol. 111-D, no. 3, pp. 188-194, 1991. H. G. Kim, S. K. Sul, and M. H. Park, “Optimal efficiency drive of a current source inverter fed induction motor by flux control,” IEEEE Trans. Industry A p p h t . , vol. 20, no. 6, pp. 1453-1459, 1984. F. Khater, R. D. Lorenz, D. W. Novotny, and K. Tang, “The selection of flux level in field oriented induction machine controller with consideration of magnetic saturation effects,” in IEEE/IAS Annu. Meet. Conf. Record, pp. 124-131, 1986. R. D. Lorenz and S. M. Yang, “AC induction servo sizing for motion and control applications via loss minimizing real-time flux control,” in IEEE/IAS Annu. Meet. Conf. Record, pp. 612-616, 1989. K. Matsuse and H. Kubota, “Digital control scheme of rotor flux in induction motor with the deadbeat response,” in IEEE/IAS Annu. Meet. Conf. Record, pp. 222-226, 1987. K. Matsuse and H. Kubota, “Deadbeat response of flux control of induction motor,” in Proc. European Conference on Power Electronics and Applications, vol. 2, pp. 895-898, 1987. K. Matsuse and H. Kubota, “Deadbeat flux level control of high power saturated induction servo motor using rotor flux observer,” in IEEE/IAS Annu. Meet. Conf. Record, pp. 409-414, 1991. 0. Ojo and M. Vipin, “Steady state performance evaluation of saturated field oriented induction motors,” in IEEE/IAS Annu. Meet. Conf. Record, pp. 51-60, 1990. T. Mizuno, J. Takayama, T. Ichioka, and M. Terashima, “Decoupling control method of induction motors taking stator core loss into consideration,” in Proc. IPEC (Tokyo, Japan), 1990, pp. 69-74. H. Kubota and K. Matsuse, “Compensation for core loss of adaptive flux observer-based field oriented induction motor drives,” in IECON’92 (IEEE), pp. 67-71, 1992.

Lipei Huang was bom in Jiangsu, China, on July 12, 1946. He received the B.E. and M.E. degrees in electrical engineering from Tsinghua University, Beijing, China, in 1970 and 1982, respectively. In 1970 he joined the Department of Electrical Engineering at Tsinghua University. Since 1990, he has been an Associate Professor of Electrical Engineering at Tsinghua University, where he has been vice-chairman of the DeDartment of Electrical Eneineerine since 1992. In 1987 he was a Visithg Scholar of Electrical kgineeryng at Tokyo Institute of Technology, Tokyo, for three months, and at Meiji University, Kawasaki, Japan, for nine months. He joined the research projects of K. Matsuse Laboratory at the Department of Electrical Engineering, Meiji University, Kawasaki, Japan, as Visiting Professor in 1993. His research interests are in power electronics, adjustable-speed drives and ac machines. He has published over 30 technical articles in the field of power electronics, wind energy utilization, and ac machines.

Yasuki Tadokoro was born in Kanagawa, Japan, on June 14, 1968. He received the B.E. and M.E. degrees in electrical engineering from Meiji University, Tokyo, Japan, in 1991, and 1993 respectively. He joined Ebara Corporation in 1993, where he has been engaged in the design and development of inverters for pumps. He is a member of the Institute of Electrical Engineers of Japan.

Kouki Matsuse (SM’88) was born in Tsingtao, China, on August 6, 1943. He received the B.E., M.E., and Ph.D. degrees in electrical engineering from Meiji University, Tokyo, Japan, in 1966, 1968, and 1971, respectively. In 1971 he joined the faculty at Meiji University as a Lecturer of Electrical Engineering. From 1974 to 1979, he was an Associate Professor at Meiji University. Since 1979, he has been a Professor in the Department of Electrical Engineering at Meiii University. In 1980 he was a Visiting Professor of Elekrical Engineering at Iowa-State University for five months. His research interests are in power electronics, microprocessor-based controllers for static power converters and drives, adjustable-speed drives, electrical linear actuators, and ac machines. Dr. Matsuse is the author of more than 100 technical articles in the field of power electronics, ac drives, and ac machines, and he holds three U.S. patents. He received the Outstanding Paper Award in 1992 from the Institute of Electrical Engineers of Japan. He is a member of the IEEE-IAS Industrial Drives Committee, the IAS Industrial Power Converter Committee, and the Japanese National Committee of IEC-TC22. He is a member of the Institute of Electrical Engineers of Japan and the Society of Instrument and Control Engineers of Japan.