byGE Bicycle Model - Enet - Main Page - …gurkan/Bicycle Model Tutorial.pdfGurkan Erdogan, Ph.D....

9

Click here to load reader

Transcript of byGE Bicycle Model - Enet - Main Page - …gurkan/Bicycle Model Tutorial.pdfGurkan Erdogan, Ph.D....

Page 1: byGE Bicycle Model - Enet - Main Page - …gurkan/Bicycle Model Tutorial.pdfGurkan Erdogan, Ph.D. VEHICLE PLANAR DYNAMICS – BICYCLE MODEL Assumptions • 2-DoF, o Lateral, y (measured

Gurkan Erdogan, Ph.D.

VEHICLE PLANAR DYNAMICS – BICYCLE MODEL

Assumptions

• 2-DoF,

o Lateral, y (measured from instantaneous center of rotation O)

o Yaw, ψ (wrt Global Axis)

• Longitudinal velocity xv is assumed to be constant.

• Small slip angles, i.e. tires operate in the linear region.

• No rear wheel steering.

• No aligning moment in both tires.

• No road gradient or bank angle.

• There are only two wheels, one in the front and one in the rear.

• No lateral and longitudinal load transfer

• No rolling and pitching motion

• No chassis or suspension compliance effects

Notations

• Lateral Acceleration of CoG in the G frame ya (calculated)

• Lateral Acceleration of CoG in the B frame. y&& (assumed to be measured)

• Yaw Rate ψ& (assumed to be measured)

• Longitudinal Velocity of Vehicle xv (assumed to be known)

• Front/Rear Tire Cornering Stiffness rf CC αα , (assumed to be known)

• Front/Rear Tire Slip Angle rf αα , (calculated)

• Front Steering Angle fδ (assumed to be measured)

• Front/Rear Wheel Velocity Angle vrvf θθ , (calculated)

• Distance from CoG to Rear/Front Axels rf LL , (assumed to be known)

• Vehicle Length rf LLL += (calculated)

• Road Radius R

Page 2: byGE Bicycle Model - Enet - Main Page - …gurkan/Bicycle Model Tutorial.pdfGurkan Erdogan, Ph.D. VEHICLE PLANAR DYNAMICS – BICYCLE MODEL Assumptions • 2-DoF, o Lateral, y (measured

Gurkan Erdogan, Ph.D.

System Model

• Lateral Force Equilibrium yryfy FFma +=

• Moment Equilibrium yrryffz FlFlI −=ψ&&

Lateral

Acceleration

in the G

frame

ψ&&&xy vya +=

Rear Tire Front Tire

Lateral

Forces in

the B frame rcryr FF δcos= fcfyf FF δcos=

Lateral

Forces in

the W frame rrcr CF αα= ffcf CF αα=

Slip Angle vrrr θδα −= vfff θδα −=

Velocity

Angle

−= −

x

rvr

v

Ly ψθ

&&1tan

+= −

x

f

vfv

Ly ψθ

&&1tan

Exact Lateral Force Balance

( ) ( )

( ) f

x

f

ffr

x

rrrx

fvfffrvrrr

fffrrr

fcfrcr

yfyry

v

LyC

v

LyCvym

CC

CC

FF

FFma

δψ

δδψ

δψ

δθδδθδ

δαδα

δδ

αα

αα

αα

costancostan

coscos

coscos

coscos

11

+−+

−−=+

−+−=

+=

+=

+=

−−&&&&

&&&

Page 3: byGE Bicycle Model - Enet - Main Page - …gurkan/Bicycle Model Tutorial.pdfGurkan Erdogan, Ph.D. VEHICLE PLANAR DYNAMICS – BICYCLE MODEL Assumptions • 2-DoF, o Lateral, y (measured

Gurkan Erdogan, Ph.D.

Approximate Lateral Force Balance

( )

( ) ff

x

ffrr

x

fr

x

ff

x

f

f

x

rr

x

f

x

r

ffr

x

f

f

x

rr

x

f

x

r

x

f

f

x

fff

x

rr

x

rrr

x

f

ff

x

rrrx

Cv

LCLCy

v

CCvym

Cv

LC

v

LCy

vC

vC

CCv

LC

v

LC

v

yC

v

yC

v

LC

v

yCC

v

LC

v

yCC

v

LyC

v

LyCvym

δψψ

δψ

δψψ

ψδ

ψδ

ψδ

ψδψ

ααααα

ααααα

αααααα

αααααα

αα

+

−+

+−=+

+

−−+

−−=

++−−

−−−=

−−+

−−−=

+−+

−−≅+

&&&&&

&&

&&&&

&&&&

&&&&&&&

11

0

f

f

x

x

ffrr

x

fr

m

Cv

mv

LCLCy

mv

CCy δψ ααααα +

−+

+−= &&&&

Approximate Moment Balance

f

x

f

ffr

x

rrrz L

v

LyCL

v

LyCI

+−+

−−−≅

ψδ

ψδψ αα

&&&&&&

State Space Representation (Jazar Chapter 10 page 612 - Equation 10.184)

022 =

+

+−

−−+

=

rf

z

ff

f

xz

ffrr

xz

ffrr

x

x

ffrr

x

fr

I

LCm

C

y

vI

CLCL

vI

CLCL

vmv

CLCL

mv

CC

yδδ

ψψ α

α

αααα

αααα

&

&

&&

&&

Centrifugal force and tire forces also have a longitudinal component in the BODY axis.

Page 4: byGE Bicycle Model - Enet - Main Page - …gurkan/Bicycle Model Tutorial.pdfGurkan Erdogan, Ph.D. VEHICLE PLANAR DYNAMICS – BICYCLE MODEL Assumptions • 2-DoF, o Lateral, y (measured

Gurkan Erdogan, Ph.D.

x

y

x

yxy

x

yxxy

x

y

dt

d

&

&&

&

&&&&

&

&&&&&&

&

&&

=

−=

−=

=

2

2

0

β

The Other Version:

f

z

ff

x

f

xz

ffrr

z

ffrr

x

ffrr

x

fr

I

LC

mv

C

vI

CLCL

I

CLCL

mv

CLCL

mv

CC

δψβ

ψβ

α

α

αααα

αααα

+

+−

−−+

=

&&&

&

22

21

ψβββ && −+−= ˆcosˆsinˆ

v

a

v

a yx

Page 5: byGE Bicycle Model - Enet - Main Page - …gurkan/Bicycle Model Tutorial.pdfGurkan Erdogan, Ph.D. VEHICLE PLANAR DYNAMICS – BICYCLE MODEL Assumptions • 2-DoF, o Lateral, y (measured

Gurkan Erdogan, Ph.D.

Steady State Steering Angle

rffssR

Lααδδ −+≅= GEOMETRY

• yx

yryf maR

vmFF ==+

2

CONSTANT Centripetal/Lateral Acc.

• 0==− ψ&&zyrryff IFLFL ZERO Angular Acc.

• yr

yf aL

LmF = y

f

yr aL

LmF = CONSTANT

• L

ma

C

L

C

F y

f

r

f

yf

f

αα

α == L

ma

C

L

C

F y

r

f

r

yr

r

αα

α == CONSTANT

yv

y

f

f

r

rss

aKR

L

L

ma

C

L

C

L

R

L

+=

−+=

αα

δ

• Under-steer gradient L

m

C

L

C

LK

f

f

r

rv

−=

αα

o Under Steer rf

f

f

r

rv

C

L

C

LK αα

αα

>>> 0

o Neutral Steer rf

r

r

f

f

vC

L

C

LK αα

αα

=== 0

o Over Steer rf

f

f

r

rv

C

L

C

LK αα

αα

<<< 0

Page 6: byGE Bicycle Model - Enet - Main Page - …gurkan/Bicycle Model Tutorial.pdfGurkan Erdogan, Ph.D. VEHICLE PLANAR DYNAMICS – BICYCLE MODEL Assumptions • 2-DoF, o Lateral, y (measured

Gurkan Erdogan, Ph.D.

Vehicle Side Slip Angle Estimation

2

ˆ

x

yxxy

x

y

dt

d

&

&&&&&&

&

&&

−=

This looks okay, but we would like to use sensor measurements.

( )

( )

( )

( )

ψββ

ψβψ

βψ

ψβψ

ββψ

β

ψβψββψβ

ψβψβψββ

ψββψββ

ψψββ

ββ

β

&

&&&&&&&&&

&&&&&&&&&

&&&&

&&&

&&&&&&&

&&&&&&

&&&&&&

&&&&

&&&

&&&

&&&&&&

&

&&&&&&&

&

&&&&&&&

−+−=

−+

+−

−=

−+++−=

−+++−=

−+++−=

−+++−=

−++−=

−=

−=

−=

−≅

−=

cossin

cossin

coscossinsin

coscossinsin

cossincossin

cossincossin

cossin

sincos1

1

1

ˆ

22

22

22

2

22

2

2

v

a

v

a

v

xy

v

yx

v

x

v

y

v

y

v

x

v

y

v

x

v

y

v

x

v

y

v

x

v

y

v

x

xyv

v

yx

v

xy

v

yxxyv

x

yxxy

v

x

x

yxxy

yx

Now we can use yaw rate and acceleration sensors to estimate the yaw rate as below.

ψβββ && −+−= ˆcosˆsinˆ

v

a

v

a yx

Maximum Vehicle Side Slip Angle

It is a saturating function of the characteristic velocity.

( ) ( )

2max

12

2

213

3

21max 32

kvv

kv

vkk

v

vkkvv

chx

ch

x

ch

xchx

=⇒>

+−−−=⇒<

β

β

Page 7: byGE Bicycle Model - Enet - Main Page - …gurkan/Bicycle Model Tutorial.pdfGurkan Erdogan, Ph.D. VEHICLE PLANAR DYNAMICS – BICYCLE MODEL Assumptions • 2-DoF, o Lateral, y (measured

Gurkan Erdogan, Ph.D.

Yaw Rate Estimation

This is actually the steady state response, i.e. the gain, of the transfer function between

the steering input and the yaw rate output.

+

=

2

2

1ch

x

f

x

v

vL

ψ&

Derivation of the above equation and the Characteristic Velocity

ψψ

δ

&&xv

x

xxv

x

x

xv

yvss

vKv

L

R

vvK

R

v

v

L

R

vK

R

L

aKR

L

+=

+=

+=

+=

2

+

=

+

=

+=

+=

2

22

11ch

x

ssx

xv

x

ss

xvx

xx

ss

xv

x

ss

v

vL

v

L

vK

v

L

vKL

v

v

L

v

L

vKv

L

δδψ

δψ

δψ

&

&

&

v

chK

Lv =2

Maximum Yaw Rate

( )( )

( )βψ

βψ

ββψ

ββ

ˆsin1

sin10

sincos

sincos

max

max

max

max

xmeasrad

x

xxmeasrad

xx

xymeasrad

aav

ava

avy

aaa

−=

++≅

++=

+=

&

&

&&&

We assume that

o CoG draws a perfect circle during a maneuver.

o The radial acceleration is in the direction of the radial force acting on the CoG.

o The radial acceleration is measured as the car travels around a circular trajectory.

o The longitudinal acceleration of the car measured along the longitudinal axis of

the vehicle also contributes to the radial acceleration of the CoG.

o All the acceleration is spent for the yaw motion, no lateral translation exist which

also means that the CoG stays on the circular trajectory.

o The max radial acceleration corresponds to a maximum yaw rate.

Note that radial and lateral accelerometers do not point at the same direction.

Page 8: byGE Bicycle Model - Enet - Main Page - …gurkan/Bicycle Model Tutorial.pdfGurkan Erdogan, Ph.D. VEHICLE PLANAR DYNAMICS – BICYCLE MODEL Assumptions • 2-DoF, o Lateral, y (measured

Gurkan Erdogan, Ph.D.

3 tane yerde yaw rate hesapladik

1.Bicycle model steady state

2.Max yawrate Steering Pad deneylerinden

1 ile 2 yi karistirip gaini bulduk sonra bicycle modelin steering input/yaw rate output

(gaini 1 olan) transfer fonksiyonundan gecirdik

(bir adet daha saturation function var en son islem olarak)

3.4W modelden gelen gercege daha yakin yaw rate var birde

Page 9: byGE Bicycle Model - Enet - Main Page - …gurkan/Bicycle Model Tutorial.pdfGurkan Erdogan, Ph.D. VEHICLE PLANAR DYNAMICS – BICYCLE MODEL Assumptions • 2-DoF, o Lateral, y (measured

Gurkan Erdogan, Ph.D.

clear all

close all clc % What is the sampling time in [s]? dt = 0.1; % [sec] % What is the longitudinal velocity of the vehicle CoG in [m/s]? vx = 10; % [m/s] % What are the vehicle parameters? % Alfa Romeo Parameter Set Caf = 42200; % [N/rad] Car = 28567; % [N/rad] Lf = 1.18; % [m] Lr = 1.52; % [m] m = 1582; % [kg] Iz = 2430; % [kgm^2]

function [Kv,sysd] = bicycle(Caf,Car,Lf,Lr,m,Iz,vx,dt) a11 = -(Caf+Car)/(m*vx); a12 = -(Lf*Caf-Lr*Car)/(m*vx^2)-1; a21 = -(Lf*Caf-Lr*Car)/Iz; a22 = -(Lf^2*Caf+Lr^2*Car)/(Iz*vx); Ac = [a11 a12; a21 a22]; b11 = Caf/(m*vx); b21 = Lf*Caf/Iz; Bc = [b11; b21]; Cc = eye(2); Dc = zeros(2,1); [Ad,Bd,Cd,Dd] = c2dm(Ac,Bc,Cc,Dc,dt); sysd = ss(Ad,Bd,Cd,Dd); set(sysd,'Name','BICYCLE MODEL') set(sysd,'InputName',{'delta_f'}) set(sysd,'StateName',{'beta','psi_dot'}) set(sysd,'OutputName',{'beta','psi_dot'}) set(sysd,'Ts',dt) Kv = m*Lr/(Lr+Lf)/Caf-m*Lf/(Lr+Lf)/Car;