a readhesion control method without speed sensor for electric railway vehicles.pdf

8/10/2019 A readhesion control method without speed sensor for electric railway vehicles.pdf

http://slidepdf.com/reader/full/a-readhesion-control-method-without-speed-sensor-for-electric-railway-vehiclespdf 1/6

A

Readhesion Control Method without Speed Sensor for Electric Railway Vehicles

Michihiro Yamashita

Railway TechnicalResearch Institute

2-8-38, Hikari,

Kokubunji

Tokyo 185-8540

JAPAN

[email protected]

Abstract

-

Railway vehides have tractive force by rolling contact between

wheel and rail. When the iriction coeiiicient between wheel

and

rail is small,

here are some cases where

wheel

slips or shids cause

running impossible. Therefore, wheel slips or skids are now

actuallystudied.

Recently, electric vehicles

fed

by inverters, and traction motors

vector control have come into wide we,and vector control without

speed sensor is at a trial stage for commercial roins.

Conventional readhesion control methods have made use of

speed-sensor (pulse generator) signal.. When vector control

without speed sensor

isused, t is

desirable not to use speed sensors

for raadhesion control, and

to

maintain unchanged performance.

We propose a method

of

slip detection without speed sensor by

focusing on the current

of

each motor. The effeetivenesr of the

proposed method

was

verified

by

simulation

raulis

and running

t a b using Shinkansen train.

1.

INTRODUCTION

Since the advent of railway vehicle with three-phase drive

in

the 1970s, power semiconductors used in electric railway

vehicles have been changed &om thyristors

to

GTO

thyristors

and

IGBTs. As

for traction motor control, vector control

has

come into wide in use in place of slip frequency control and

speed

sensorless vector control is in trial stage for commercial

trains[l],

[Z].

Anti-slip and readhesion control of railway

vehicles

has

also

taken

a great progress by utilizing wheel axle

speed

informatio631,[4],[5],[6],[~,[8].

So,

futnre

locomotives

and electric multiplennitswith speed sensorless vector control

should have at least the same adhesion performance

as

altematives with speed sensors.

At first characteristics of wheel slip phenomenon and anti-slip

control for railway vehicles are

mumerated

and

running

test

results that show basics of anti-slip readhesion control with

speed sensors

are

explained.

We proposed the novel anti-slip readhesion control without

speed sensor191, [lo].

By

this

method, small wheel slip can be detected and the

performance of

t h i s

control is expected to

be as

high

as the

control with speed

sensors. After tbis,

an outline of the novel

anti-slip readhesion control without speed sensor is

proposed,

and

the

effectiveness of that

is

verified by simulation and

running

test

results using Shinkansen.

U.CHARACrWsncsOFWHEEL sm

PHENOMENONAND

m

SLIP CONTROL

Tomoki Watanabe

Railway Technical

Research

Institute

2-8-38,

Hikari, Kokubunji

Tokyo 185-8540

JAPAN

Tomwoia@rhi. rjp

I When wheels adhere to the rail, the difference between the

wheel peripheral

speed

and the vehicle speed (i.e. slip velocity)

is negligibly small under

normal

conditions.

2)

Slip phenomenon changes depending on the status of the

contact surface of the wheel and the rail.

3) The number of pulses per rotation of speed sensor is

normally between

40

and

90

that is much smaller

than

that of

encoders that are widely used in other industrial applications.

So,

in addition to pulse numher counting at certain intervals,

pulse width measurement is needed to obtain as accurate speed

as possible.

4

ormally, a

speed

sensor is mounted on the traction motor.

Vertical movement of wheel axle causes rotational displacement

of the traction motor that results in shifting of the speed sensor

pulses and fluctuationof the calculated velocity. Soappropriate

averaging is necessary for wheel axle spee detection.

5) One inverter

feeds

several motors (Multiple Motor Drive)

or one motor (Individual Motor Drive). Multiple Motor Drive is

less expensive

than

Individual Motor Drive. But the wheel

diameter difference causes current imbalance among motors

in

a

group,

so

that it is necessary to

keep

he difference as low as a

few millimeters.

In.OLITLJNEOFApsn-SLIP READHESION

CONTORL WITHOW

SPEED ENSOR

Considering

our

experience[9],

[lo],

we propose the

following anti-slip readhesion control without speed sensor.

I

We focus on multiple motor drive that is widely used for

E M U S .

2)

We detect currents of each traction motors. Normally, the

inverter output c m t i.e. summation of motor currents) is

used for inverter control for multiple motor drive.

3)

As for tbe wheel axle acceleration alternativeby which slip

is detected, we use time differential of tractionmotor rotational

speed that is derived &om the inverter hquency and slip

bquency. If the motor flux is constant, the slip eequency is

proportional to the q-axle component of the motor current. So,

in vector control, we can obtain the slip frequency of each

traction motor &om relevant motor current.

4) As

for the slip velocity alternative, by which slip is

detected, we propose to use the current difference of traction

motors. The difference due

to

imbalance of wheel diameters

or

motor characteristics can

be

compensated in time of motor field

excitation or adhesion

run

for example.

5 The adhesion force at the

point

of slip detection

can be

deduced &ommotor torque and motor rotational acceleration.

They can

be

obtained

rom the vector controller.

0-7803-78 17-2/03/ 17.00,@2003IEEE

.~

29I



IV. SIMULATIONMODEL AND CONTROL

In

this chapter, we explain a simple case what one inverter

feeds

two

traction motors.

A.

Motor

Drive

Figure I shows

a

block diagram of the multiple induction

motors drive system.

I &-

Fig. 1

Block

dia- of

the

multiple

induction

motor

drive

system

The

state equations of induction motor and drive system on

the

d-q

co-ordinates used for simulation are as follows.

(3)

M 2 M 2

LZZ LILZ

=ri+-r2, a= l - -

where

i,:Stator current vector;

v,:Stator

voltage vector; A:Rotor

fluxvector; w,:Primaryangular velocity; o,:Rotational angular

velocity;

o,

Slip angular velocity;

0 :

nduction motor torque;

r

:Wheel radius (0.43111);

G

:Gear ratio(3.01);P Number ofpole

pairs 2); R,:Stator resistance; R,:Rotor resistance; L,:Primary

self inductance; &:Secondary self inductance; M Mutual

inductance;

J:

Moment

of

inertia ofa drive system;

51:

The

load

torque corresponds to

the

adhesion;

-n

:

n means a number of

induction motor;

The torque T < s expressed as follows.

(4)

L2

L

= p -

( Zdilq -

Zdi ld)

B.

Motor Current

Figure 2 shows the diagram of tbe relation between the

voltage and current vectors in adhesion drive. The phases and

amplitudes

of

primary-current vector

il-l

and

i,,

supplied

to

each induction motor are identical.

Figure 3 shows the behaviour of stator current vector without

constant current control when the second wheel slips. The stator

current i _lf IM-1 holds the same value. On he other hand, the

stator current

i,,

of IM-2

in

Figure

2

becomes

i,.;

in Fig. 3, and

the slip angular velocity

o -2

of IM-2 and torque current

i,,,

decrease due to slip,

so

the torque of IM-2 re-%ecreases too.

When

the constant current control

of

the vector controller is

active, the stator current vector i , holds

the

same value (Fig. 4).

The current difference vector, Ai holds the same value as in the

case of Fig. 3, because the flux change is negligible and there is

the

same electrical slip difference between two, motors. That

means the torque current i,,,

of IM-I

increases and the current

il,, of

IM-2

decreases. The torque current difference

i,,,-i,,,

is

almost similar to that of Fig. 3.

C.

Relafion befween Speed Difference

and

Current Difference

This

paragraph describes the relation between the speed

The speed difference and current difference are defined

as

difference and current difference or each motor.

follows.

Speed difference

AK =

K-z -

K - i

5 )

Currentdifference N = i i p _ i - i t q _ z 6)

The

torque component current

i , ,

and

the flux

component

current ,, are constantly controlled by tbe vector control method.

The slip frequencyAm

of

each motor when a slip occurs is

calculated by using the following equation of approximation.

where

Awr

is

defined as the calculated slip frequency difference.

The axle speed difference AF l onmbelow is calculated by the

torque component current of each motoras follows.

3.6r

PG

AVt=-Aub

292



D.

Slip

Detection wifhoufSpeed Sensor

Since the torque current of

each

induction motor behaves as

mentioned above, a threshold is set in the torque current

differencesbetween

two

motors to detect slips. Figure

5

sbows

the diagram of slip detection. Regarding the wheel acceleration,

we use the motor speeds derived from the vector con ~ollerand

the same detection method as the one commonly used in the

conventional anti-slip readhesion control

with

speed sensor.

E.

Adhesion

Presumption

We propose a new approach to presume adhesion force

between wheel

and

rail without speed sensor. The load torque

equivalent to adhesion torque is expressed by the following

e&atioo from the state equation (I ) of the chive system

(9)

The torque r,, is calculated by the following approximate

expression, where

6

s a reference value

of

rotor

flux

inkage.

Not to

use a speed sensor, d q . 2/dt s presumed from (1 1).

The slip angular velocity of u1-2 is expressed by (12) in

adhesion run.

The flux b2ardly changes when the 2nd wheel starts

slipping. Theref&e, the differentiation of o,-~s approximated

by

W.

Substitute (13)into (1 1) and some manipulation of

(9)

lead to

the load torque that is equivalent to the presumed value of

decreased adhesion.

It

is possible

to

recover the torque current on the basis of the

presumed value after readhesion.

VI q

VI

t

Fig. 3 Vestordiagrams wbm

a slip

OCNFJ.

Withan c n t a t m mL ' Ibe 2nd ha17

s

st slip)

293



V I

T

200

Fig.4Vest01

diagramswhen

lip

w m .

Withonstantm t control.

The

2nd shafl is at slip)

I W

I

.

.......

I IC

coqara te r

i i L i

r

q -

Slip detectionsignal

i i q 2 t-i q 2 f

comarater

I l L I

r

q -

Slip detectionsignal

Fig.5 Method to detect slipsethos speed s e

v. SlMUL4TlON

In

this

chapter, we perform simulation to verify the

effectiveness

of

the proposed anti-slip readhesion control

method with

a

two-motor

(IM-1

and

IM-2)

drive system fed by

one inverter.

In

case of the slip detection system with sped

sensor, a speed difference of

0.81r

(between the vehicle

speed and the speed of slipping wheel) and the wheel

acceleration of W s re set as thresholds. Normally the

threshold value of O S k is the limit of the conventional anti-

slip readhesion control with speed sensor. In case of the

proposed control without speed sensor, slips are detected by the

torque current difference

of 30A

between

the

two motors. In the

simulation, adhesion i.e. load to que was suddenly

decreased

when

the time was 0.lsec. The

amount of

adhesion

drop is

10%in

Figs 6 ,7 and

8.

Figure

6

shows the simulation results of torque and current in

case of conventional anti-slip readhesion control with

speed

sensor. It is difficult to detect

small

slips by the conventional

control method with speed sensor (Fig.

9).

Therefore, the torque

reel

f adhesion axis

(IM-I)

increases, which will cause all-axle

slip.

Figure 7 shows the simulation

results

when a slip is detected

without speed sensor. By the proposed method of slip detection

based on the current difference, slip was detected

very

quickly.

This

time we assume that adhesion drop of the

wheel

axle 2

294

remains.

So

the axle2 slips again when torque is recovered

after

readhesion. From

OUT

experience of running tests with anti-slip

control without adhesion presumption, slip-stick repetition as

shown in Fig.

I

occurs when adhesion force between wheel and

rail is much

lower

than tractive force[7].

30

O D 0.5 I

o 1.5 2.0

TimC (roe)

1 600

1.400

P

E

l,wo

0600

b

4W

am

1W

6m

l W

0

a

I <

..........

0.0 0.5 I I5 2.0

Tm.l .)

Fig.6Responseswhen a slip occu~s

(10%

decrease

in

adhesion

force,

wt

speeds )

0.0

0.5

1.0 1.5 2.0

rims

sec)

1m

E 1.w



Figure 8 shows the characteristics of readhesion control with

adhesion force presumption. The presumed adhesion force is

almost the Same

as

the decreased adhesion, good presumption of

adhesion force isperformed.

Figure 7 and 8 prove the effectiveness of the anti-slip

readhesion control without speed sensor.

VI.RUNNING

ESTS

RESULTS

In

this chapter, the relation between wheel peripheral speed

and motor current for each motor was described in running test

Figure 9 shows the test electric multiple-unit (e.m.u). We

sprayed

water on

the rail in

front

of the axle No. 4 of the vehicle

for measurement Figure IO shows a block diagram of the

tractionmotor circuit.

An

nverter with vector control

feeds

four

traction motors. Figure 11 shows the tes t results of a nnming

test of slips. Slips occur at three sections,from I7sec to ZOsec,

from

22sec

to 28sec and from 33sec to 38sec.

In

thispaper, we

analyze the sectionwhich does not detect slips between I7sec

and ZOsec. In that

section,

he axleNo.4 starts slip first because

of sprayed

water

in

front,

and then otber wheels slip. As

described in the above section, the toque of the slipping axles

decreases and that of the adheringaxles increases.

Figure

12

shows the characteristic between cwe nt difference

and speed difference, in which there is a linear relationship.

Equation

(14)

is derived by substituting parameters

of

traction

motor for (8). The experimental values as shown in Fig. 12

agree

well

with the theoretical values that derived from (14).

using

shinkansen.

Avt'(1- 4) =0.022

( i iq_ i -i ir .4)

(14)

295

Tberefare, even if a speed sensor s not used, a threshold for slip

detection can be set up 6 o m the n t difference and motor

parametem

12

17 22 27 32 37 42

1-1

i w

I . ~

i

, a e c , , . h

12

17

22

27

.

32 37 42

Fig.11

T c ~ tesultsof a d g u t when slipsm m .



ACKNOWLEDGMENl

1.0 I

153.0 i

J.5 I

Fig.

12

Liner sharactnistitieb e e n urrent difference and spe difference

1.0 r 290

288

Fig.13 Behavior ofspeed difference

m d

mm d i t T m c e between he I‘ axle

and

4’ axleswhen slipocnns. (from 1 6 s o

2 2 s

Figure

13

shows he relation of the speed difference

AVt(l-4)

calculated

from

PG sensors and the speed difference AVf’ 14)

estimated fi.om the cnrrent difference. The speed difference

AVr’( l4) agrees well withthe speed difference

AVt 1-4).

From

this characteristic, it is expected that not only the speed

differences

hut

also accelerations for slip detection can be

estimated satisfactorily.

VI.

CONCLUSION

This

paper proposed a novel anti-slip readhesion control

method without spe-ed sensor for electric railway vehicles wt

multiple induction motor drive, as summarized below.

I) It is possible to detect slips by tinding the torque current

difference

of

each induction motor.

2)

The proposed method makes it easier to detect small slips

than conventional method with speed sensors.

3) Simulation and running test results show the effectiveness

of the proposed anti-slip readhesion control without speed

sensor.

This study was supported by the Program for Promoting

Fundamental Transport Technology Research. The authors wish

to

thank

the Corpo&on for Advanced Transport Technology

(CAlT).

REFERENCES

[I]

KYuld, A.Ujiie et

al.,

“DevelopmentofHighPerfamanss

Traction

Drive

System

without Speed

Sensors”, in Proceedings of 20W lntemational

Pow= Elemon ics C onference (IPEC-Tokyo2000).pp.1373-1378

KKondw,

K Y u h et al., ‘Corm1ChractRislicsof

an

Induction

Motor

Speed

Sensorless

Conml

System

for the

Railway

Vehicle

Traaion”, in

Roceedbgs of

2002

Power

Conversion Conference-Osaka (PCC-

OSAKAZ002), pp.861-866

T.Watanabe, M.0- “Improvement

of nadbesive

chnraCterirtics

of

e l d c motor vehicles

by

Slip Velocity Feedback.Tarque CnnuaL

Sp&” ’ in Proceedings

of

the

1992

Symposium

on Power

Elatmaier,

Eleshical

Drives,

Advanced

Electrical

Motors (Speedam),pp.373-378

T.Watanabe,M.0-

‘Xealirationof

Anti-slip/sIide Control in

Railway

Motor

Vehicles by

Slip

Velocity

Feedback

Torque C o n m Y ~ n

Proceedings

nd applications EPE93),Vol. 6,pplS6-161

[ 5 ]

T.Watanabe,

MOpasa,

S.Ohe,

“Improvement of Rc-adhesivs

Characteristics

of

E l d c Motor vehicles,” in

Roceeding of he

1993

ht-tional Conference

on

speedup

Technology

far

Railway

and

Maglev

Vehicles (STECH’93). pp.243-247

T.Watanabe, S.Ohc,

M.0ga.q ‘Tmctical Methai of Readhesion control

f Electric mot^ Vehicles

wt

Three Phase Drives”,

in Rocadings

of

6* EloweanC o n f m c c onPowerEleomnics and AwlicatimWE‘95).

[2]

[3]

[4]

[6]

V o l 2 , pp 4 4 8 4 5 2

T

Watanabe. M

Yamanaka. “Adhaion Phmomma

and

0pm”Jhm of

Readhenon

Cmml

for HI Swcd ShlnLansen

Tram

wth whec l -h l

[7]

Adhesion Rediction,” in

hoc;edmg

of World Cong cso on

Railway

Research 1997 (WCRR’97), Vol. D,pp.359-364

[8]

T.Watanabc,“Application

of

Anti-slip R d c s i m Cone01

with

Adhesion

Prediction

to

Caamutu Electric

Multiple

Units,” in Roceedinp of

international Conference

Railway

Traction

Systems

(RTSOl)

[9]

T.Watanabc, M.Y-hiR “ A Novel Anti-slip C m m l

withoar S p e d

S e ”

f~ E l d c

Railway

Vehicles”,inRoceedingofIECON2Wl

[IO] T.Watanabe, M.Yamaohita, ‘‘Baric-Shdy f Anti-slip Conml

without

Speed

SeosOr f~

Multiple

Motor Drive ofEle&ic

Railway

Vehicles”, in

Rosadings

of

2 2

POrVcr convmion C o n f m n C o O ~(Pcc-

OSAKA2002).

pp.IO2b1032

296

a readhesion control method without speed sensor for electric railway vehicles.pdf

Documents