0 Votos favoráveis0 Votos desfavoráveis

79 visualizações6 páginasOct 24, 2008

© Attribution Non-Commercial (BY-NC)

PDF, TXT ou leia online no Scribd

Attribution Non-Commercial (BY-NC)

79 visualizações

Attribution Non-Commercial (BY-NC)

Você está na página 1de 6

Rotor Resistance and Speed Based on the Feedforward Torque Control Scheme

S. K. Jeong', Z. G. Lee', H. A. Toliya?, P.Niazi'

'Division of Mechanical Engineering, F'ukyong National University, Pusan, Korea

E-mail : skjeong@pknu.ac.!u

2

Department of Electrical Engineering, Texas A&M University, College Station, TX 77843-3 128

E-mail : toliyat@ee.tamu.edu

Abstract - In tbis paper, a new speed sensorless induction regulators. Third, the electromagnetic torque is controlled

motor scheme which can work at any speed including the lem very fast and independent of the rotor flux without inducing

speed is presented. The proposed method is robust to rotor spike in the currents because the control voltage was derived

resistance variations. Simultaneous on-line estimations of speed theoretically from the steady state currents values. Since

and rotor resistance are realized based on a feedforward type

torque control scheme. The rotor flux with a low frequency speed and rotor resistance estimations do not depend on the

sinusoidal waveform is used to help the estimation. The control stator resistances, stator voltages, and any other derivatives of

scheme has no current minor loop to determine voltage currents, it is possible to achieve speed sensorless with good

references. Since the proposed estimation does not depend on performance even during low speed region. Especially, we do

any derivative term of currents and stator voltages, it not inject any high frequency signal to estimate the rotor

contributes to speed sensorless with good performance at very resistance, thus we expect less current ripples and we do not

low speed repion. Furthermore, the proposed control b simply need to design a high band-pass filter to reject them out.

using motor parameters and stator curreuts without Hence, we can reduce the burden of hardware to inject high

determining any PI gains for current feedback and any signal frequency signals and also software to detect these high

injection for the rotor resistance estimation. Simulations results

snpported by experiments are given to show the effectiveness of

fiequency signals.

this method. In this paper, at 6rsf the feedfonvard type torque control is

described in detail using the machine equivalent circuit

I. INTRODUCTION eqnations. Then,the estimation eqnations for speed and rotor

resistance are derived from stator currents in d, q reference

A number of sensorless induction motor drives have been frame. Atter that, the speed sensorless based on the torque

developed in the past [1-51. However, the performances of control scheme is descriied- Finally, through several

these sensorless drives are not sufficient when compared to simulation results using a PWM voltage-source inverter and

the sensored ones. A major drawback of these drives is due to experiments based on a DSP control system, the validity of

the difficulty of control at low stator frequency. During this the proposed method is verified.

region of opmtiou, the signal-to-noise ratio of the stator

current is decreased significantly and the effect of stator D. THE MOTOR MODEL AND TI33 FEEDFORWARD TORQUE

resistance voltage drop is not negligible. Another difficulty is CONTORI.

to ensure the robustness of drives against parameter

variations, especially rotor resistance. Also, it is well known Suppose that the stator current is x, ,the rotor current is x2,

that simultaneous estimation of speed and rotor resistance is and the control input stator voltage is U,, then the dynamic

hardly obtained in the vector control induction motor drives equations of induction motor(I.M) are given as follow:

with constant rotor flux [3].

To solve these problems, we suggest a novel speed

sensorless scheme based on a feedfonvard torque control

technique [6]. The control scheme has some different features

compared with the classical vector controls. First, it controls

[;I=[ MRi DI -+ ide)

W R2 + hMD p'].

( D - j d e ) x2

(1)

rotor flux as a sinusoidal waveform in the d,q referencz where R,, L,, I , ( i 4 2 : statcu and mtor) are the

h m e without affecting the torque control performance. It is induct?ncedI*in~qrespedively..Also,M isthe

possible to estimate on-line both speed and rotor resistance mutual induclance, 8, is the eleceical rmgularpositios m d D is the

using the rotor flux.Second, it does not have any current feed

back loops. Thus, we do not need to consider any phase Metmtial ope" (4~It 3 is)n.o t e d h e bat xI,x,, and U,

compensations based on the delay between stator voltages me o o m p h variables which have two "pmds a and B , m

and stator currents. Since control voltage can be determined thestaticmaxyrefermceftame.

using this feedfonvard technique, we can exclude Ibe instaataneouS elech"gne6c tcque of LM using the

complicated process in order to design PI gains in c m t variable5x, andx, is given by

theoretically we can control the instantaneous torque without

T=NMlm[xl ?,I (2) having any transient torque.

The command voltage as a control input which can realize

Where N is the pole pair number. The symbol “Im” very quick torque response is derived from the stator side of

rqmsents the imaginary part of the variables and the the induction motor equation (1) using these xIand x l .

horizontal bar “-” means conjugate complex variable.

Generally, the rotor flux and the rotor current can be

defined as rotating vectors which have amplitudes and phases

given by (3) and (4).

In (3) and (4), the f (1) and g(t) represent phase angles of

the rotor flux and the rotor current, respectively. They will be m. SPEED SENSOWSS CONTROL .ALGORITHM

abbreviated as f and g hereafter to simplify the equations.

Since the electrical torque of LM can be defined as a vector The rotor speed can be estimated using (I 1) taking the

product of the rotor current and the rotor flux such as imaginary part of (6) and replacing the torque term by the

T = N;krsin(-g), a new expression for x2 is obtained from stator current according to (9).

(4).

Substituting (3) and (5) in the rotor side of the induction It is noted that the second term of the right hand side in

motor equation (l), we can drive (6). (11) m e a slip frequency. As a result, the speed can be

calculated very simply using the motor parameters, stator

frequency, the rotor f l u , and the stator current. From (ll),

the estimated speed is closely related to the motor parameters.

Particularly, it is well known that the rotor resistance varies

during motor operation due to the increase of temperame in

From the real part of (€9,the inverse tangent of g is written the rotor winding. Therefore, we need to estimate its value.

as. From the real part of (9), the rotor resistance can be

described as (12).

(7)

by substituting(7) in (5).

It is noted here again that the rotor flux has to be

commanded as a sinusoidal waveform to estimate the rotor

resistance, since the deferential value of rotor flux is

necess;uy in the dq reference fiame. Another i m p ” t point

is that the rotor resistance has a possibility of diversion if the

The stator current x1 is also rewritten by substituting for denominator of (12) is near zero. To overcome this problem,

4 and the newly obtained x, in (3). the rotor resistance is calculated by integrating the stator

current and the rotor flux at every quarter cycle such as (13).

terms about time t. Thus, it is obvious from (2) that

1838

Fig. 1 illustrates the suggested speed sensorless control complicated thanthe previous stepwise command.

system The system is consisted of four major parts, speed Motor speed is set at 15 rpm and IS0 rpm,respectively.

controller, torqne controller, speed estimator, and rotor The rotor flux is commanded to be a consfant value in Fig.

resistance estimator. Z(a). However, in Fig. 2@), it varies with time which is given

In this control scheme, there is no current feedback loop by 1, =O.SZ5{l+O.lsin(lOf)} Weber.

and also the current information is not used for producing the Fig. Z(a) and Fig. Z@) are consisted of three and four

voltage command pieces of figures, respectively. The tirst one represents torque

responses according to the step and M torque references,

N. NUMERICAL

sIMvLAnoN respectively. They sho-w that the torque responses accurately

To verify the proposed sensorless control scheme, three follow their respected commanded values. The second one

separate simulations are performed using the P W M voltage- expresses voltage commands to achieve the torque references.

source inverter and an 1.M whose parameters are listed in When the toque reference is changed abruptly, a very high

Table I. The carrier fkquency is set at 4 kHz and the voltage is added to the steady state value to eliminate the

sampling times for speed calculation and current integration traosient torque response based on the f e e d f m d torque

of 1.M model are set at 1 ms, and 1 11s respectively. At lirst, control. The third one represents the stator currents responses

the simulation results for feedfomard torque control under the voltage commands. The last one in Fig. Z(b) depicts

performances are investigatedbased on Fig. 1. the variable rotor flux. These results show that the torque

controller performs correctly even though the rotor flux is

commanded to be sinusoidal.

Fig. 3 shows the estimation results of the speed and rotor

PR I.S[!dq N 2 resistance. Among them, the first one depiqs the estimation

speed and rotor resistance, respectively. These results show

uR 220/380[P7 xR 6.6/3.8[,4] that the estimated values well coincide with their actual

values. The rotor flux waveform was given as

RI 3.7I[Q] R, 2.79[Q]

1 ,=0.525(1 tO.lsin(f)) Weber. The initial value of the rotor

L, 236.79[ mH ] L, 236.79[ mH ] resistance in Fig. 3@) was given appropriately to confirm

M 220.42[mH] J o.0O21Nm~s2/r~d1 the convergence characteristic of it to its actual value.

V. EXPEIUMENTALREsULT

Fig. Z(a) shows the torque response when the torque

Fig.4 shows the experimental hardware set up based on

reference is changed abmptly stepwise f “ 3 N.m to 5 N.m

at time equal 50 milliseconds. Fig. 2@)also shows the torque DSP system The dead time of inverter was 5 fis and the

carrier ftequency was &t at 2.5 kHz,and the sampliig time

rcsponse to the commanded torque, M, whicb is more

1839

was 1 ms. Also the A/D converter to calculate the current had

12 bits resolution and 3.2 ps conversion time. The rotor flux

was givenas .I,=0.6(1+0.1sin(t)) Weberduringexperiment.

Fig. 5 represents experimental results of estimating the

speed and the rotor resistance using the hardware shown in

Fig. 4. The motor parametem are shown m Table I. Fig. 5(a)

shows, the reference speed, actual speed from an encoder,

and the estimated speed, under steady state at 3 rpm. The

estimated speed shows good agreement with the actual speed.

Fig. 5 @ ) shows an estimated rotor resistance using (12).

Actually there are some disagreements between the real value

and the estimated one. However, the filtered estimated value

Fig.4. DSP -based h d w n e sct up for the expnimCnt followed the a d value very closely after a few seconds.

1840

U.. .

g 0.4

”i IJ

1.0 I I

2 03

Fig. 5(c) and Fig. S(d) illustrate the stator currents in d, q It is sbownthatiheresultofspdsensorlessm m l followsfbe

axes and the mtor flux, respectively. They were used for actual speed closely.

estimating the rotor speed and rotor resistance. The ermr

between the rotor flux reference and the response in Fig. S(d) VI CONCLUSION

is the reason of imprecise estimation of the rotor resistance.

Fig. 6 shows an experimental result of speed sensorless In this paper, a novel speed sensorless control scheme

control. Since low speed operation was of interest, speed was based on the feedfonvard torque control was proposed.

commanded from 0 rpm to 3 rpm under no-load. Especially, the method aimed at the simultaneous estimation

of the rotor speed and the rotor resistance on-line. Through

simulation results, the validity of the proposed sensorless

control scheme was initially verified. The experimental

results for sensorless control with variable rotor flux

reference also showed good agreement with the theory even

at very low speed such as a few of rpm

REFERENCES

Drive With 0ecaupli.g Contml and Stability Analysis of Sped

E h a t i o n , ” IEEE Dm. I d . Applimf,,vol. 49,2002, no. 2, pp. 444-

455.

121 Y. Kinpara and M.Koyama, “Speed Sennorless V m Control Method

0.0 5.0 10 I3 20 of Induction Mota Including B Low Speed Region,” JIEE, vol. IZOD,

time 11d m 1 2oW,pp. 223.229.

Fig. 6. ExpCrimental result of the speed remorless conml. [3] K Akam a d A. Kawmura, “Sensorl.nn V q Law and Zero Speed

Enimatiau with Oa-Line Secanday Re&sJtanee Estimation of

1841

Induction Mom Without Adding Any Signal:' BEE T ? m . 1 4

+., 1999,pp. 187-193.

[4] 1.1. Ha and S.K SUI, "Sasmleas Ficld-OrientatiOn Cmtml of 811

Industion Mashine bv. Hi&-Frequency

- . . Signal . In1ectiols"EEE

. Trow.

,MA&., 1999, vol. 35, no. I, pp. 45-51.

[SI €L Kubora and K. MataLse, "Sped Senrmless Field-Onmted Cmml

of lndustion M m a witb Rotor Rerisla~ceAdqWion,'' IEEE Trow.

MA&-., vol. 30,no. 5, l994,pp. 1219-1224.

[6] T. Hayashi, Y. Fujii md T. Sekiguchi, "Sfudy on an Analytical

Solution for h s " w t m Torque h t m l of an Induction Motor:'

. h d & $ c f E E , vol. 4,1989, pp. 323-324.

[7] K. Ohnishi, N.Ma&, and Y. Hori, "Estimation,Idcatification, and

S-les Contml in Motion Conwol System," proceedings of I=,

vol. 82.m. 8,Aupst 1994,pp.1253-1265.

I842

## Muito mais do que documentos

Descubra tudo o que o Scribd tem a oferecer, incluindo livros e audiolivros de grandes editoras.

Cancele quando quiser.