2. Mathematical Description of Systemsocw.snu.ac.kr/sites/default/files/NOTE/9203.pdfPerception and...
Transcript of 2. Mathematical Description of Systemsocw.snu.ac.kr/sites/default/files/NOTE/9203.pdfPerception and...
Perception and Intelligence LaboratorySchool of Electrical Engineering at SNU
2. Mathematical Description of Systems
Linear System Representation
Causality and Lumpedness
Linear Systems
Linear Time-Invariant (LTI) Systems
Linearization
Examples of Linear Systems
Discrete-Time Systems
1Linear Systems
Perception and Intelligence LaboratorySchool of Electrical Engineering at SNU
Linear System Representation
Differential Equation
Impulse Response
Transfer Function
State Space Equation
2Linear Systems
0
( ) ( , ) ( )
y( ) G( ) u( )
( ) ( ) ( ) ( ) ( )( ) ( ) ( ) ( ) ( )
t
ty t G t u d
s s s
x t A t x t B t u ty t C t x t D t u t
τ τ τ=
=
= += +
∫
&
( ) ( ) ( 1) ( ) ( 1)( ) ( ( ), ( ),..., ( ), ( ), ( ),..., ( ))n n n n ny t f y t y t y t u t u t u t− −=
Perception and Intelligence LaboratorySchool of Electrical Engineering at SNU
Linear System Representation
Definition 2.1: The state of a system at time is the information at that, together with the input , for , determines uniquely the output for all .
3Linear Systems
0( )x t 0t0t ( )u t 0t t≥
0t t≥( )y t
Voltage Source
( )u t ( )y tLi CvOutput Voltage
0 0
0 0
1 0 2 0
( ), ( )
( ), ( )
( ), ( )
L Ci t v t
y t y t
x t x t
&
Example
( )y t ( )u t0t
can be uniquely determined for any input if initial values of induction current and capacitor voltage at
State:
Perception and Intelligence LaboratorySchool of Electrical Engineering at SNU
Causality and Lumpedness
Causal System (Nonanticipatory System)The current output depends on only the past and current inputs but not on the future inputs
4Linear Systems
Lumped System (Finite Dimensional System)The number of state variables is finiteIf infinite, Distributed System
Voltage Source
( )u t ( )y tLi CvOutput Voltage
Perception and Intelligence LaboratorySchool of Electrical Engineering at SNU
Linear Systems
Linear System Satisfying superposition principle: additivity + homogeneity
5Linear Systems
00
0
1 1 0 2 2 01 1 0 2 2 0 0
1 1 0 2 2 0 0
( )For ( ), , 1, 2
( ),
( ) ( )then ( ) ( ),
( ) ( ),
ii
i
x ty t t t i
u t t t
x t x ty t y t t t
u t u t t tα α
α αα α
⎫⇒ ≥ =⎬≥ ⎭
+ ⎫⇒ + ≥⎬+ ≥ ⎭
Perception and Intelligence LaboratorySchool of Electrical Engineering at SNU
Linear Systems
Zero-input Response
6Linear Systems
00
0
( )( ),
( ) 0, zi
x ty t t t
u t t t⎫⇒ ≥⎬= ≥ ⎭
Zero-state Response
00
0
( ) 0( ),
( ), zs
x ty t t t
u t t t= ⎫
⇒ ≥⎬≥ ⎭
00
0
( )( ) ( ),
( ), zi zs
x ty t y t t t
u t t t⎫⇒ + ≥⎬≥ ⎭
By additivity
Response = Zero-input Response+Zero-state Response
Perception and Intelligence LaboratorySchool of Electrical Engineering at SNU
Linear Systems
Input-Output DescriptionAssume initial state is zero.Define piecewise continuous function of input:
7Linear Systems
( ) ( ) ( )
where0,
( ) 1/ ,
0,
i ii
i
i i i
i
u t u t t t
t t
t t t t t
t t
δ
δ
Δ
Δ
≈ − Δ
⎧ <⎪
− = Δ ≤ < + Δ⎨⎪ ≥ + Δ⎩
∑ ( )u t
t
( )iu t
Δ
1( ) ( ) ( ) ( )i i i iu t t t u t u tδΔ − Δ = Δ =Δ
Perception and Intelligence LaboratorySchool of Electrical Engineering at SNU
Linear Systems
Input-Output Description
8Linear Systems
Let be the output for the input , i.e. ( , )ig t tΔ ( )it tδΔ −
( ) ( , )i it t g t tδΔ Δ− →
By homogeneity
( ) ( ) ( , ) ( )i i i it t u t g t t u tδΔ Δ− Δ → Δ
By additivity
( ) ( ) ( , ) ( )
( ) ( )
i i i ii i
t t u t g t t u t
u t y t
δΔ Δ− Δ → Δ
≈ ≈
∑ ∑
Perception and Intelligence LaboratorySchool of Electrical Engineering at SNU
Linear Systems
Input-Output Description
9Linear Systems
The output for the input
0
( ) ( , ) ( )
( ) lim ( , ) ( )
( ) ( , ) ( )
i ii
i ii
y t g t t u t
y t g t t u t
y t g t u dτ τ τ
Δ
ΔΔ→
∞
−∞
≈ Δ
= Δ
=
∑
∑
∫
( )y t ( )u t
where
( ) ( , ) : Impulse Responset g tδ τ τ− →
Perception and Intelligence LaboratorySchool of Electrical Engineering at SNU
Linear Systems
Input-Output Description
10Linear Systems
Causal
0
0
0
( ) ( , ) ( )
( , ) ( )
( , ) ( ) ( , ) ( )
( , ) ( )
t
t t
t
t
t
y t g t u d
g t u d causal
g t u d g t u d
g t u d relaxed
τ τ τ
τ τ τ
τ τ τ τ τ τ
τ τ τ
∞
−∞
−∞
−∞
=
= ⇐
= +
= ⇐
∫∫∫ ∫
∫
Relaxed at :initial state at is 0
( , ) 0, for t<g t τ τ=
0t 0t
Perception and Intelligence LaboratorySchool of Electrical Engineering at SNU
Linear Systems
Input-Output Description
11Linear Systems
MIMO System
0
( ) ( , ) ( )t
tt t dτ τ τ= ∫y G u
where
11 1
21 22
1
( , ) ... ( , )( , ) ( , ) ... ...
( , )... ...( , ) ... ( , )
p
q qp
g t g tg t g t
t
g t g t
τ ττ τ
τ
τ τ
⎡ ⎤⎢ ⎥⎢ ⎥=⎢ ⎥⎢ ⎥⎢ ⎥⎣ ⎦
G
Perception and Intelligence LaboratorySchool of Electrical Engineering at SNU
Linear Time Invariant Systems
12Linear Systems
Linear Time Invariant (LTI)A system is said to be time invariant if
and any , we have
00
0
( )( ),
( ),x t
y t t tu t t t
⎫⇒ ≥⎬≥ ⎭
T
00
0
( )( ),
( ),x t T
y t T t t Tu t T t t T
+ ⎫⇒ − ≥ +⎬− ≥ + ⎭
Perception and Intelligence LaboratorySchool of Electrical Engineering at SNU
Linear Time Invariant Systems
13Linear Systems
If the system is LTI,
( , ) ( , )( , 0) (let =- )( )
g t g t T Tg t Tg t
τ ττ τ
τ
= + +
= −
≡ −
Output of LTI system
0 0( ) ( ) ( ) ( ) ( )
t ty t g t u d g u t dτ τ τ τ τ τ= − = −∫ ∫
Perception and Intelligence LaboratorySchool of Electrical Engineering at SNU
Linear Time Invariant Systems
14Linear Systems
Example) unity-feedback system2 3
3
1
( ) ( 1) ( 2) ( 3) ...
( 3)i
g t a t a t a t
a t
δ δ δ
δ∞
=
= − + − + − +
= −∑Output
0 01
1
( ) ( ) ( ) ( ) ( )
( )
t ti
i
i
i
y t g t u d a t i u d
a u t i
τ τ τ δ τ τ τ∞
=
∞
=
= − = − −
= −
∑∫ ∫
∑Unit time delay
a( )u t ( )y t
Perception and Intelligence LaboratorySchool of Electrical Engineering at SNU
Linear Systems
Transfer Function
15Linear Systems
Laplace transform of
s
0 0
s
0 0
s s
0 0
y(s) ( ( ) ( ) )
( ( ) ( ) )
( ) ( ) ,
y(s) g(s) u(s)
t t
t
v
g t u d e dt relaxed
g t u d e dt causality
g v e dv u e d v t t vτ
τ τ τ
τ τ τ
τ τ τ τ
∞ −
∞ ∞ −
∞ ∞− −
= − ⇐
= − ⇐
= ⇐ = − = +
=
∫ ∫∫ ∫∫ ∫
s
0
g(s) : ( )
g(s) : ( ) ( ) t
g t
g g t e dt∞ −= = ∫L
(s) (s) (s)=y G uTransfer Function Matrix: MIMO case
Perception and Intelligence LaboratorySchool of Electrical Engineering at SNU
Linear Systems
Properness of Transfer Function
16Linear Systems
g(s) ( ) / ( )N s D s=
g(s) proper deg ( ) deg ( )g( ) zero or constant
D s N s− ⇔ ≥⇔ ∞ =
g(s) strictly proper deg ( ) deg ( )g( ) zero
D s N s− ⇔ >⇔ ∞ =
g(s) biproper deg ( ) deg ( )g( ) non-zero constant
D s N s− ⇔ =⇔ ∞ =
g(s) improper deg ( ) deg ( )| g( ) |
D s N s− ⇔ <⇔ ∞ = ∞
Perception and Intelligence LaboratorySchool of Electrical Engineering at SNU
Linear Systems
Properness of Transfer Function Matrix
17Linear Systems
(s) is (strictly) proper if all entries are (strictly) proper−G
1
(s) is biproper if (s) is square and
both (s) and (s) are proper−
−GG
G G
Perception and Intelligence LaboratorySchool of Electrical Engineering at SNU
Linear System
State Space Equation
18Linear Systems
( ) ( ) ( )( ) ( ) ( )
x t Ax t Bu ty t Cx t Du t
= += +
&
Laplace Transform
sx(s) - x(0) x(s) u(s)y(s) x(s) u(s)
A BC D
= += +
Which implies1 1
1 1
x(s) (sI ) x(0) (sI ) u(s)y(s) (sI ) x(0) (sI ) u(s) u(s)
A A BC A C A B D
− −
− −
= − + −
= − + − +
Transfer Function Matrix1(s) (sI )C A B D−= − +G
Perception and Intelligence LaboratorySchool of Electrical Engineering at SNU
Linearization
Nonlinear System
19Linear Systems
( ) ( ( ), ( ), )( ) ( ( ), ( ), )
x t f x t u t ty t h x t u t t
==
&
Linearization at an operating point 0 0( ), ( )x t u t
0 0( ) ( ) ( ), ( ) ( ) ( )x t x t x t u t u t u t= + = +
0 0 0
0 00 0
( ) ( ) ( ) ( , , )
( , , ) (.)
x t x t x t f x x u u tf ff x u t x u Ox u
= + = + +
∂ ∂= + + +
∂ ∂
&& &
0 0
( )
where = , (.)
x t Ax Buf fA x B u Ox u
= +∂ ∂
= +∂ ∂
&
⇒
Perception and Intelligence LaboratorySchool of Electrical Engineering at SNU
Implementation and Examples
Op-Amp Circuit Implementation: P. 17Figure 2.7,
20Linear Systems
Examples: (p. 18 -29)Cart with inverted pendulum, Satellite in orbit, Hydraulic tanks,RLC circuits
Perception and Intelligence LaboratorySchool of Electrical Engineering at SNU
Discrete-Time Systems
Sampling Period: T
21Linear Systems
[ ] ( ), [ ] ( )[ ] ( )
u k u kT y k y kTx k x kT
= =
=
Linear System: homogeneity, additivityResponse= Zero-state response + Zero-input response
Impulse Sequence 1 if 0
[ ]0 if 0
kk
kδ
=⎧= ⎨ ≠⎩
Perception and Intelligence LaboratorySchool of Electrical Engineering at SNU
Discrete-Time Systems
22Linear Systems
Input Sequence [ ] [ ] [ ]
mu k u m k mδ
∞
=−∞
= −∑
[ ] [ , ]k m g k mδ − →
By homogeneity[ ] [ ] [ , ] [ ]k m u m g k m u mδ − →
By additivity
Impulse Response Sequence
[ ] [ ] [ , ] [ ]m m
k m u m g k m u mδ − →∑ ∑Input-Output Description
[ ] [ , ] [ ]m
y k g k m u m= ∑0
[ ] [ ] [ ]k
my k g k m u m
=
⇒ = −∑
CausalRelaxedLTI
Perception and Intelligence LaboratorySchool of Electrical Engineering at SNU
Discrete-Time Systems
23Linear Systems
Z-Transform
0y(z) ( [ ]) [ ] k
ky k y k z
∞−
=
= = ∑Z
Discrete Transfer Function
( )
0 0
( )
0 0
0 0
y(z) ( [ ] [ ])
[ ] [ ]
[ ] [ ]
g(z) u(z)
k m m
k m
k m m
k m
l m
l m
g k m u m z z
g k m z u m z
g l z u m z
∞ ∞− − −
= =
∞ ∞− − −
= =
∞ ∞− −
= =
= −
= −
=
=
∑ ∑
∑ ∑
∑ ∑
Perception and Intelligence LaboratorySchool of Electrical Engineering at SNU
Discrete-Time Systems
24Linear Systems
Z-Transform of
[ 1] [ ] [ ] [ ] [ ][ ] [ ] [ ] [ ] [ ]
x k A k x k B k u ky k C k x k D k u k+ = +
= +
State-Space Equations
[ 1]x k +
( 1)
0 0
1
( [ 1]) [ 1] [ 1]
( [ ] [0] [0])
(x(z) [0])
k k
k k
l
l
x k x k z z x k z
z x l z x x
z x
∞ ∞− − +
= =
∞−
=
+ = + = +
= + −
= −
∑ ∑
∑
Z
Perception and Intelligence LaboratorySchool of Electrical Engineering at SNU
Discrete-Time Systems
25Linear Systems
Z-Transform of State-Space Equations
1 1
1 1
x( ) [0] x( ) u( )y( ) x( ) u( )x( ) ( I ) [0] ( I ) u( )
( ) ( I ) [0] ( ( I ) ) u( )
z z zx A z B zz C z D zz z A zx z A B z
y z C z A zx C z A B D z
− −
− −
− = += +
= − + −
= − + − +
If zero initial state1( ) ( ( I ) ) u( )y z C z A B D z−= − +
Transfer Function1( ) ( I )z C z A B D−= − +G
Perception and Intelligence LaboratorySchool of Electrical Engineering at SNU
Discrete-Time Systems
26Linear Systems
Example: compound interest calculation
Interest: 0.015%[0] 1, [ ] 0, 1, 2,[ ] (1.00015) k
u u i ig k
= = =
=
Impulse Response
Output
0[ ] (1.00015) [ ]
kk m
my k u m−
=
= ∑Transfer Function
1
0 0
1
g( ) (1.00015) = (1.00015 )
1= 1 1.00015 1.00015
k k k
k kz z z
zz z
∞ ∞− −
= =
−
=
=− −
∑ ∑
Perception and Intelligence LaboratorySchool of Electrical Engineering at SNU
System Type Internal Description
External Description
Distributed, Linear(Causal, Relaxed)
Lumped, Linear(Causal, Relaxed)
Distributed, Linear,Time-invariant
Rumped, Linear,Time-invariant
Concluding Remarks
27Linear Systems
Example: compound interest calculation
( ) ( )( ) ( )
x A t x B t uy C t x D t u= += +
&
0
( ) ( , ) ( )t
ty t g t u dτ τ τ
−= ∫
0
( ) ( , ) ( )t
ty t g t u dτ τ τ
−= ∫
0( ) ( ) ( )
y(s) = G(s) u(s)
ty t g t u dτ τ τ= −∫
x Ax Buy Cx Du= += +
&0
( ) ( ) ( )y(s) = G(s) u(s)
ty t g t u dτ τ τ= −∫