Experimental Identification
Stochastic methods
Deterministic methods
Transfer function
Bode plot
Bode plot
computation
from Transfer
function
Electrical system
Cross-correlation function
Other methods of
parameterization
Numerical deconvolution
Basic terms
Geometric similarity:
the model has the same shape, keeps the shape similarity
the created model can be touched, it is a physical model
the physical model allows to realize experiments and study the properties
of the original using the same physical processes (for example the airflow
around the model of the car in the wind tunnel).
Car (real)
Mathematical model
We can define also another model, abstract mathematical model of the original
mathematical model.
Mathematical model
it is not possible to realize the experiments based on the same physical
processes,
it allows to investigate the processes of the original using their
mathematical description solution of the mathematical models.
Creation of the mathematical model has the following steps:
definition of the discovered processes,
definition of the observed symptoms definition of the system on the real
object.
Car suspension
F,x
Mechanical model
Mathematical model
d 2 x(t )
dx(t )
m
+
b
+ kx(t ) = F (t )
dt
dt 2
d 2 x(t )
dx(t )
m
+
b
+ kx(t ) = F (t )
2
dt
dt
Cybernetic similarity
System is a set of the elements and linkages between them which has
defined properties.
Surrounding of the system is a set of the elements, which are not
elements of the defined systems, but they have important relations to the
system.
Surrounding
(Environment)
P2
P4
System
P6
P5
P1
P3
Environment
Inputs
Outputs
System
vehicle vertical
motion
yaw
body
roll
wheel liftl
pitch
wheel
vehicle longitudinal
motion
steering
motion
rolling
motion
lateral motion
Driver
IDI
Break
brzdov
pedal
pedl
plyn.
pedl
Gas pedal
rychl.st.
gear
Subsystm:
Subsystem:
Horizontal
HorizonzlnDynamics
dynamika
Subsystm:
Subsystem:
Longitudinal
Dynamics
Podln dynamika
Longitudinal motion
podln
pohyb
Wheel
rotation
nerovnosti
Quality
of the
vozovky
road
Longitudinal
acceleration and
podln zrychlen
deceleration
a zpodn
Subsystm:
Subsystem:
Cross
Dynamics
Pn
dynamika
lateral motion
bon pohyb
naten, naklnn
Subsystm:
Subsystem:
Vertical
Dynamics
Svisl dynamika
Vertical motion of the
pohyby karoserie
wheels
oten kol
velocity resistance
odpor in
rychlost
curve
v zatkch
Steeringvolant
wheel
psoben
wind vtru
Lateral
acceleration
pn
zrychlen
naklpn
sklpn
Road
Traffic on the
road
Side wind
Surface of the road
Quality of the road
Steering:
Steering wheel,
gearshift, break
pedal
Goal of the
trip
Driver
Car position
Car
Goal of the
trip
Side wind
Surface of the road
Quality of the road
Steering:
Steering wheel,
gearshift, break
pedal
Driver
Car position
Car
Action
Control
System
Selected state variables
Car position, velocity, direction of the movement
M
Fx
Fg
FN
u = r
Motion equation of the car
mv! = Fx
Motion equation of the rotating wheel
m
g
v
J
M
Fx
FN
r
J! = rFx M
Fx = Fz
Fz = mg
Fx = mg ,
Friction coefficient
Fx
=
Fz
Slip
The friction coefficient between the surface and the wheel (tyre) depends on
the slip .
1,2
such asfalt
dry asphalt
Friction coefficient
souinitel
brzd.
Break
moment
momentum
r
Fx
Fz
asfalt
asphalt
0,8
wet asphalt
0,6
mokr
0,4
snh
ice
led
snow
0,2
0
20
40
slip
60
skluz
80 100 %
vu
u
= 1
v
v
-1/mk
1/J
s
breaking
path
Velocity v
1
Mux
break
momentum M(t)
final
Wheel
momentum inertia
Angular
velocity w
v, u
lambda
f(u)
mk*g
s
Clock
Mux
lambda =1-u/v
mi=f(lambda)
Fx
Simulation results
Breaking momentum
Breaking path
Deceleration
Simulation results
Breaking momentum
Breaking path
Deceleration
-1/mk
r
deceleration
Model ABS
s
velocity v
0.19
0.01s+1
Slip error
valve
ABS
Hydraulic
system
dynamics
x
Mux
v, u
lambda
f(u)
mk*g
s
Angular
velocityt w
100
w lambda
s
Breaking path
1/J
final
inertia
momentu
m
On the wheel
1
s
tlak
Mux
lambda =1-u/v
mi=f(lambda)
Fx
6
break
momentum
Slip
Static systm
description only using the
static characteristic:
y = f ( u)
u1
u
2
u =
!
ur
y1
y
2
y =
!
y l
x! = f (x, u)
x(0) = x 0
y = g( x, u )
Dynamic system non linear, t-variant
x! = f (x, u, t ),
y = g(x, u, t ).
x(0) = x 0
x! = A (t )x(t ) + B (t )u(t )
x(0) = x 0
x! = Ax(t ) + Bu(t )
y = Cx(t ) + Du(t )
x(0) = x 0
x! = Ax(t ) + bu(t )
y = c T x(t ) + du(t )
x(0) = x 0
Transfer Function
Transfer function
bm s m +!+b1 s + b0
G ( s) =
a n s n +!+a 2 s 2 + a1 s + a 0
bm ( j ) m +!+b1 j + b0
G ( j ) =
a n ( j ) n +!+a 2 ( j ) 2 + a1 ( j ) + a 0
Z-Transfer function
Y ( z ) b0 + b1 z 1 + ! + bm z m
G( z ) =
=
U ( z ) 1 + a1 z 1 + ! + a n z n
1
y (n ) + an 1 y (n 1) + ... + a1 y! + a0 y = b0u
The first state variable is equal to the output y:
The second and next variable
is defined as a derivative of
the previous:
x! i = xi +1,
x! n =
x1 = y
i = 1, 2, ..., n 1
d ny
dt n
y (n ) = a0 y a1 y! ... an 1 y (n 1) + b0u
The time derivative of the last
state variable xn can be
expressed from the given
x! n
differential equation:
()
diy
( ) = xi +1,
dt i
i = 1, 2, ..., n 1
Initial condition
The initial conditions- initial values of the state variables x1(0),
xn(0) are equal to the initial values of the output variables y(0).
This can be obtained from the definition formulas for the state
variable for t=0.
xi (0) = y (i 1) (0), i = 1,2,..., n
x1 (0) = y (0)
x2 (0) = y" (0)
!
xn (0) = y ( n 1) (0)
x! 1
x! 2
=
=
.
.
x! i
=
=
=
=
=
.
.
x! n
x2
x3
x i +1
= a 0x1 a1x 2
a 2 x3
a n 1x n + b0u
y = x1
x!1 $ ! 0
& #
x!2 & # 0
& #
. & # .
. & # .
&=#
x!i & # .
. && ## .
. & # .
& #
x!n % " a0
0
.
.
a1 a2
a3
1 .
1
an1
$!
&#
&#
&#
&#
&#
&#
&#
&#
&#
&#
&#
%"
x1 $ !
& #
x2 & #
& #
. & #
. & #
&+#
xi & #
. && ##
. & #
& #
xn % "
$
0 &
0 &
&
. &
. &
& u
0 &
. &&
. &
&
b0 %
x1
x
2
.
.
y = 1 0 0 . . . 0
x i
.
.
x n
If the derivatives of the input occur on the right side of the differential
equation:
y (n ) + an 1 y (n 1) + ... + a1 y! + a0 y = bmu (m ) + bm 1u (m 1) + ... + b1u! + b0u
x = x1
1
0
0
1
a1
0
0
x1 0
x 0
2
. . .
+ u
0 . .
1 . .
a n 1 x n 1
The state variables x1, x2, ..., xn represent the partial outputs of the system for the input
signal u and his derivatives u,u etc.. The right side of the given differential equation is a
linear combination of the input signal u and its derivatives. The state variables represent
the outputs of the system for the inputs x1 u, x2 u, x3 u, etc.
x! 1
x! 2
=
=
.
.
x! i
=
=
=
=
=
.
.
x! n
x2
x3
xi
= a 0 x1 a1x2
y = b0x1 + b1x2
a 2 x3 . .
a n 1x n + u
+ b2 x3 +. . + b mx m+1
x1 (0) = y(0), x 2 (0) = y! (0), ..., x n (0) = y (n1) (0), u(0), u! (0), ..., u (m1) (0)
!
#
#
#
#
#
#
#
#
#
#
#
"
x!1 $ ! 0
& #
x!2 & # 0
& #
. & # .
. & # .
&=#
x!i & # .
. && ## .
. & # .
& #
x!n % " a0
0
.
.
a1 a2
a3
1 .
1
an1
$!
&#
&#
&#
&#
&#
&#
&#
&#
&#
&#
&#
%"
x1 $
& !
x2 & #
& #
. & #
. & #
&+#
xi & #
. && ##
. & #
& "
xn %
0
0
.
.
0
.
.
1
$
&
&
&
&
& u
&
&
&
&
%
x1
x
2
.
.
b1 . . b m . 0 0
xi
.
.
x n
y = b0
b2
b1
xm+10
xn
....
an-1
x20
xm+1
....
am
....
x3
a2
....
b0
x10
x2
a1
x2
x1=x
a0
Y( s)
U( s)
Seriel programming
G( s) =
s n i
s p i
G i ( s) =
1
s p i
(s n1)(s n2 ) ... (s - n m )
(s p1)(s p2 ) ... (s - p n )
G( s) = G i ( s)
i =1
1
p - pn
Yi ( s)
s ni
=
Yi +1( s) s p i
yn
yi+1
....
p - ni
p - pi
yi
y2
....
p - n1
p - p1
x! i = p i x i + (p i n i )y i +1
y i = x i + y i +1
y=y1
x! i = p i x i + y i +1
yi = ni xi + pi xi + yi
!#"#
$
xi
yi+1
xi0
pi - ni
yi+1
xi0
xi
yi
xi
xi
-ni
-pi
pi
yi
Ai
b
+ n
s pi a n
i =1
G( s) =
A i = lim (s pi )G(s)
s p i
The state variables are defined x1, x2, ..., xn using the partial transfer functions:
Xi (s)
Ai
=
,
U(s) s pi
i = 1, 2, ..., n
x! i = pi xi + A i u,
Output equation:
i = 1, 2, ..., n
n
b
Y(s) = Xi (s) + n Us
an
i =1
b
y = x1 + x 2 + ... + x n + n u
an
Parallel structure
Matrix form:
p1 0
0 p
2
x! = .
.
.
0
.
.
.
.
.
.
.
.
.
.
pi
.
.
.
.
0
A1
A
0
.
.
. x + A i u
.
.
.
.
A
p n
n
b
y = 1 1 1 . . 1 . . 1 x+ n u
an
The obtained form is also called Jordan form of the state equations.
A1
x1
x1
bn
an
p1
.
.
.
xi0
u
Ai
xi
xi
pi
.
.
.
xn0
An
xn
pn
xn
pi = + j,
pi +1 = j.
A i = A + Bj,
A i +1 = A Bj.
A1
A + Bj
A - Bj
A
+ ... +
+
+ ... + n
s pi
s- p n
s- ( + j) s- ( j)
x! i = xi + xi +1 + 2Au
x!i +1 = xi + xi +1 + 2Bu
y i = xi
Parallel programming
Matrix form of
the state
model with
complex eigen
values:
(all matrix
elements are
real numbers)
x! 1 p1
. .
. .
. .
x! i .
=
x! i +1 .
. .
. .
. .
x! n 0
.
.
.
.
.
.
.
.
0 x1 A1
. . .
. . .
. . .
. xi A i
.
+
u
. xi +1 A i +1
. . .
. . .
. . .
p n x n A n
x1
.
.
x i
y = 1 . . . 1 0 . . . 1
x i +1
.
.
.
x n
1 1
3
G( s) =
s s+1 s+ 4
Solution:
u
3
s+4
x3
1
s+1
x2
1
s
y = x1
State variables:
1
X1(s) = X2 (s)
s
1
X2 (s) =
X (s)
s+1 3
3
X3(s) =
U(s)
s+ 4
x! 1 = x2
x! 2 = x2 + x3
x! 3 =
- 4x3 + 3u
x! 1 0
1
0 x1 0
!
x
=
0
1
1
2
x2 + 0 u
x! 3 0
0 4 x3 3
G(s) =
s2 + 5s + 6
You have to create the state model using the parallel combination of the
transfer functions (parallel programming).
G( s) =
6
1
3
+
+
5s 2( s + 1) 10( s + 5)
6
U (s )
5s
1
X 2 (s ) =
U (s )
2(s + 1)
3
X 3 (s ) =
U (s )
10(s + 5)
X 1 (s ) =
s s2 + 6s + 5
Parallel programing
5
u
6
1
x! 2 = x2
u
2
3
x! 3 =
5x3 + u
10
x! 1 =
y = x1 + x2 + x3
5
x! 1 0
0
0 x1 6
1
!
0 x2 + u
x2 = 0 1
2
x! 3 0
0 3 x3 3
10
x1
y = 1 1 1 x2
x3
1
1
1
0,5
1
1,2
y1 y!1 y! 2
y2 + u1 + u!1
12
4
6
12
12
12
2,4
1
1
y! 2 =
y1 y2 u2
5
5
5
!y!1 =
1
1
1
0,5
1
1
y!1 y10 = ( y1 y10 ) + (u1 u10 ) ( y2 y20 ) + y1
y2 + u1 d
4
10
6
12
12
12
0
x! 1 =
1
1
1
y1 + u1 y2 + x1
4
10
6
x10 =
1
0,5
1
y1
y2 + u1
12
12
12
1
1
1
y10 u10 + y20 + y!10
4
10
6
1
1
1
y1 y10 = y1 + u1 y2 + x1 d
4
10
6
1
1
1
x! 2 = y1 + u1 y2 + x1
4
10
6
x20 = y10
y1 = x 2
1
1
2,4
y2 y20 =
y1 y2 u2 d
5
5
5
0
x! 3 =
2,4
1
1
y1 y2 u2
5
5
5
y2 = x3
x30 = y20
1
0,5
1
y1
y2 + u1
12
12
12
1
1
1
x! 2 = y1 + u1 y2 + x1
4
10
6
2,4
1
1
x! 3 =
y1 y2 u2
5
5
5
x! 1 =
State variables:
1
0,5
1
x2
x3 + u1
12
12
12
1
1
1
x! 2 = x1 x2 x3 + u1
4
6
10
2,4
1
1
x! 3 =
x2 x3
u2
5
5
5
x! 1 =
State equations:
Matrix form:
x! 1
x! 2 =
x! 3
0
1
0
1
12
1
4
2,4
5
x
y1 0 1 0 1
y = 0 0 1 x2
2
x3
0,5
1
12 x1
5
u1
1 1
x2 +
0
6
10
u2
1 x3
1
5
5
Initial conditions:
1
1
1
y10 + y0 u0 + y20
4
10
6
x20 = y10
x30 = y20
x10 =
-1
24
Block diagram:
-1
5
x10
x1
-1
12
u1
1
x20
1
10
-1
4
12
25
u2
-1
5
-1
6
x30
-1
5
x2
x3
(n )
Initial values:
y( t 0 ) = y 0 ,
y( t 0 ) = y0 ,
.
.
.
x! = f ( x, u, t )
( n 1)
y( n 1) ( t 0 ) = y 0
,
y = g( x, u, t )
Definition of
the state
variables:
x1 = y ,
x2 = y! ,
.
.
.
xn = y (n 1) ,
Definition
of the
state
equations:
x!1 = x2 ,
x!2 = x3 ,
.
.
.
x!n 1 = xn .
x! n = f (x1, x2 , ..., xn , u, t )
x! 1 = x2 ,
x! 2 = x3 ,
Initial values:
x! 3 = x 4 ,
.
x1(t 0 ) = y0 ,
x2 (t 0 ) = y0 ,
.
.
.
.
x! n = f ( x1, x2 , ..., x n , u , t )
Output equation:
x n (t 0 ) = y0n 1.
y = x1
i = 1, 2, ..., n -1