Plan of the Lecture

I Review: control, feedback, etc.

I Today’s topic: state-space models of systems; linearization
Plan of the Lecture

I Review: control, feedback, etc.

I Today’s topic: state-space models of systems; linearization

Goal: a general framework that encompasses all examples of

interest. Once we have mastered this framework, we can
proceed to analysis and then to design.
Plan of the Lecture

I Review: control, feedback, etc.

I Today’s topic: state-space models of systems; linearization

Goal: a general framework that encompasses all examples of

interest. Once we have mastered this framework, we can
proceed to analysis and then to design.

Reading: FPE, Sections 1.1, 1.2, 2.1–2.4, 7.2, 9.2.1.

Chapter 2 has lots of cool examples of system models!!
Notation Reminder

We will be looking at dynamic systems whose evolution in

time is described by di↵erential equations with external
inputs.

We will not write the time variable t explicitly, so we use

x instead of x(t)
dx
ẋ instead of x0 (t) or
dt
d2 x
ẍ instead of x00 (t) or 2
dt
etc.
Ach
i
=
Example 1: Mass-Spring System
x

m u
Example 1: Mass-Spring System
x

m u

Newton’s second law (translational motion):

F
|{z} = ma
total force
Example 1: Mass-Spring System
x

m u

Newton’s second law (translational motion):

F
|{z} = ma = spring force + friction + external force
total force
Example 1: Mass-Spring System
x

m u

Newton’s second law (translational motion):

F
|{z} = ma = spring force + friction + external force
total force

spring force = kx (Hooke’s law)

friction force = ⇢ẋ (Stokes’ law — linear drag, only an approximation!!)
Example 1: Mass-Spring System
x

m u

Newton’s second law (translational motion):

F
|{z} = ma = spring force + friction + external force
total force

spring force = kx (Hooke’s law)

friction force = ⇢ẋ (Stokes’ law — linear drag, only an approximation!!)
F = kx ⇢ẋ + u
Example 1: Mass-Spring System
x

m u

Newton’s second law (translational motion):

F
|{z} = ma = spring force + friction + external force
total force

spring force = kx (Hooke’s law)

friction force = ⇢ẋ (Stokes’ law — linear drag, only an approximation!!)
mẍ = kx ⇢ẋ + u
Example 1: Mass-Spring System
x

m u

Newton’s second law (translational motion):

F
|{z} = ma = spring force + friction + external force
total force

spring force = kx (Hooke’s law)

friction force = ⇢ẋ (Stokes’ law — linear drag, only an approximation!!)
mẍ = kx ⇢ẋ + u
Move x, ẋ, ẍ to the LHS, u to the RHS:

mẍ + ⇢ẋ + kx = u
Example 1: Mass-Spring System
x

m u

Newton’s second law (translational motion):

F
|{z} = ma = spring force + friction + external force
total force

spring force = kx (Hooke’s law)

friction force = ⇢ẋ (Stokes’ law — linear drag, only an approximation!!)
mẍ = kx ⇢ẋ + u
Move x, ẋ, ẍ to the LHS, u to the RHS:
⇢ k u
ẍ + ẋ + x =
m m m
Example 1: Mass-Spring System
x

m u

Newton’s second law (translational motion):

F
|{z} = ma = spring force + friction + external force
total force

spring force = kx (Hooke’s law)

friction force = ⇢ẋ (Stokes’ law — linear drag, only an approximation!!)
mẍ = kx ⇢ẋ + u
Move x, ẋ, ẍ to the LHS, u to the RHS:
⇢ k u
ẍ + ẋ + x = 2nd-order linear ODE
m m m
Example 1: Mass-Spring System

m u

⇢ k u
ẍ + ẋ + x = 2nd-order linear ODE
m m m
Example 1: Mass-Spring System

m u

⇢ k u
ẍ + ẋ + x = 2nd-order linear ODE
m m m

Canonical form: convert to a system of 1st-order ODEs

Example 1: Mass-Spring System

m u

⇢ k u
ẍ + ẋ + x = 2nd-order linear ODE
m m m

Canonical form: convert to a system of 1st-order ODEs

ẋ = v (definition of velocity)
⇢ k 1
v̇ = v x+ u
m m m
Example 1: Mass-Spring System
x

m u

State-space model: express in matrix form

✓ ◆ ✓ ◆✓ ◆ ✓ ◆
ẋ 0 1 x 0
= k ⇢ + 1 u
v̇ m m v m
Example 1: Mass-Spring System
x

m u

State-space model: express in matrix form

✓ ◆ ✓ ◆✓ ◆ ✓ ◆
ẋ 0 1 x 0
= k ⇢ + 1 u
v̇ m m v m

Important: start reviewing your linear algebra now!!

I matrix-vector multiplication; eigenvalues and eigenvectors; etc.
General n-Dimensional State-Space Model
0 1 0 1
x1 u1
B C B C
state x = @ ... A 2 Rn input u = @ ... A 2 Rm
xn um
General n-Dimensional State-Space Model
0 1 0 1 4
x1 u1
B C B C
state x = @ ... A 2 Rn input u = @ ... A 2 Rm
ratix

Gestater O-ip
xn um

matrix
0 1 0 10 1
0 10 1
ẋ1 A x1 B u1
.
@ .. A = @ .
A @ .. A + @ A @ ... A
n⇥n n⇥m
ẋn matrix xn matrix um
-atsegn
of egn
General n-Dimensional State-Space Model
0 1 0 1
x1 u1
B C B C
state x = @ ... A 2 Rn input u = @ ... A 2 Rm
xn um

0 1 0 10 1 0 10 1
ẋ1 A x1 B u1
.
@ .. A = @ .
A @ .. A + @ A @ ... A
n⇥n n⇥m
ẋn matrix xn matrix um

Dr
ẋ = Ax + Bu
Partial Measurements
0 1 0 1
x1 u1
B C B C
state x = @ ... A 2 Rn input u = @ ... A 2 Rm
xn um
Partial Measurements
0 1 0 1
x1 u1
B C B C
state x = @ ... A 2 Rn input u = @ ... A 2 Rm
xn um
0 1
y1
B .. C
output y = @ . A 2 Rp y = Cx C– p ⇥ n matrix
yp
Partial Measurements
0 1 0 1
x1 u1
B C B C
state x = @ ... A 2 Rn input u = @ ... A 2 Rm
xn um
0 1
y1
B .. C
output y = @ . A 2 Rp y = Cx C– p ⇥ n matrix
yp

ẋ = Ax + Bu
y = Cx
Partial Measurements
0 1 0 1
x1 u1
B C B C
state x = @ ... A 2 Rn input u = @ ... A 2 Rm
xn um
0 1
y1
B .. C
output y = @ . A 2 Rp y = Cx C– p ⇥ n matrix
yp

ẋ = Ax + Bu
y = Cx
Example: if we only care about (or can only measure) x1 , then
0 1
x1
B x2 C
B C
y = x1 = 1 0 . . . 0 B . C
@ .. A
xn
State-Space Models: Bottom Line

ẋ = Ax + Bu
y = Cx
State-Space Models: Bottom Line

ẋ = Ax + Bu
y = Cx
State-space models are useful and convenient for
writing down system models for di↵erent types
of systems, in a unified manner.
State-Space Models: Bottom Line

ẋ = Ax + Bu
y = Cx
State-space models are useful and convenient for
writing down system models for di↵erent types
of systems, in a unified manner.

When working with state-space models, what are

states and what are inputs?
State-Space Models: Bottom Line

ẋ = Ax + Bu
y = Cx
State-space models are useful and convenient for
writing down system models for di↵erent types
of systems, in a unified manner.

When working with state-space models, what are

states and what are inputs?
— match against ẋ = Ax + Bu
Example 2: RL Circuit
+ VR
+ +
VS I VL
Example 2: RL Circuit
+ VR
+ +
VS I VL

VS + V R + VL = 0 Kirchho↵’s voltage law

VR = RI Ohm’s law

VL = LI˙ Faraday’s law

VS + RI + LI˙ = 0
Example 2: RL Circuit
+ VR
+ +
VS I VL

VS + V R + VL = 0 Kirchho↵’s voltage law

VR = RI Ohm’s law

VL = LI˙ Faraday’s law

VS + RI + LI˙ = 0

R 1
I˙ = I + VS (1st-order system)
L L
Example 2: RL Circuit
+ VR
+ +
VS I VL

VS + V R + VL = 0 Kirchho↵’s voltage law

VR = RI Ohm’s law

VL = LI˙ Faraday’s law

VS + RI + LI˙ = 0

R 1
I˙ = I + VS (1st-order system)
L L
I – state, VS – input
Example 2: RL Circuit
+ VR
+ +
VS I VL

VS + V R + VL = 0 Kirchho↵’s voltage law

VR = RI Ohm’s law

VL = LI˙ Faraday’s law

VS + RI + LI˙ = 0

R 1
I˙ = I + VS (1st-order system)
L L
I – state, VS – input
Q: How should we change the circuit in order to implement a
2nd-order system?
Example 2: RL Circuit
+ VR
+ +
VS I VL

VS + V R + VL = 0 Kirchho↵’s voltage law

VR = RI Ohm’s law

VL = LI˙ Faraday’s law

VS + RI + LI˙ = 0

R 1
I˙ = I + VS (1st-order system)
L L
I – state, VS – input
Q: How should we change the circuit in order to implement a
2nd-order system? A: Add a capacitor.
spring-Mass-Damper
///////

I ke for

41a -
-

#11 F
.

x Ma Mc
F
:
kx
=
- -

I
tait Jude Gener
 Cinci .

sE
e
*

WR

I
from kUL

I
t
iR
0 +
! fidt
=

+
L 02
-
C
-

-
-
I = da
-
-

At

R+Ld
I7
Y
=
V Hee
F
=

<>

MES L >
- (9 Ri 14
=

0 +
+

k >
Y
~ :
*
↳L-c circuit : -

Hoy
-
Delive Force to Current Analogy ·

· C
Example 3: Pendulum

✓ `
✓
Te
external
mg sin ✓ torque

mg
Example 3: Pendulum
Newton’s 2nd law (rotational motion):
✓ `
✓
Te
external
mg sin ✓ torque

mg
Example 3: Pendulum
Newton’s 2nd law (rotational motion):
✓ `
✓ T = |{z}
|{z} J ↵
|{z}
Te total moment angular
external torque of inertia acceleration
mg sin ✓ torque

mg
Example 3: Pendulum
Newton’s 2nd law (rotational motion):
✓ `
✓ T = |{z}
|{z} J ↵
|{z}
Te total moment angular
external torque of inertia acceleration
mg sin ✓ torque
= pendulum torque + external torque
mg
Example 3: Pendulum
Newton’s 2nd law (rotational motion):
✓ `
✓ T = |{z}
|{z} J ↵
|{z}
Te total moment angular
external torque of inertia acceleration
mg sin ✓ torque
= pendulum torque + external torque
mg
pendulum torque = mg sin ✓ · |{z}
`
| {z }
force lever arm
Example 3: Pendulum
Newton’s 2nd law (rotational motion):
✓ `
✓ T = |{z}
|{z} J ↵
|{z}
Te total moment angular
external torque of inertia acceleration
mg sin ✓ torque
= pendulum torque + external torque
mg
pendulum torque = mg sin ✓ · |{z}
`
| {z }
force lever arm
2
moment of inertia J = m`
Example 3: Pendulum
Newton’s 2nd law (rotational motion):
✓ `
✓ T = |{z}
|{z} J ↵
|{z}
Te total moment angular
external torque of inertia acceleration
mg sin ✓ torque
= pendulum torque + external torque
mg
pendulum torque = mg sin ✓ · |{z}
`
| {z }
force lever arm
2
moment of inertia J = m`

mg` sin ✓ + Te = m`2 ✓¨

Example 3: Pendulum
Newton’s 2nd law (rotational motion):
✓ `
✓ T = |{z}
|{z} J ↵
|{z}
Te total moment angular
external torque of inertia acceleration
mg sin ✓ torque
= pendulum torque + external torque
mg
pendulum torque = mg sin ✓ · |{z}
`
| {z }
force lever arm
2
moment of inertia J = m`

mg` sin ✓ + Te = m`2 ✓¨

g 1
✓¨ = sin ✓ + Te (nonlinear equation)
` m`2
Example 3: Pendulum
g 1
✓¨ = sin ✓ + Te (nonlinear equation)
` m`2
Example 3: Pendulum
g 1
✓¨ = sin ✓ + Te (nonlinear equation)
` m`2
For small ✓, use the approximation sin ✓ ⇡ ✓
Example 3: Pendulum
g 1
✓¨ = sin ✓ + Te (nonlinear equation)
` m`2
For small ✓, use the approximation sin ✓ ⇡ ✓
3 g 1
sinHqL ✓¨ = ✓+ Te
2 ` m`2
q 1

-3 -2 -1 1 2 3
-1

-2

-3
Example 3: Pendulum
g 1
✓¨ = sin ✓ + Te (nonlinear equation)
` m`2
For small ✓, use the approximation sin ✓ ⇡ ✓
3 g 1
sinHqL ✓¨ = ✓+ Te
2 ` m`2
q 1
State-space form: ✓1 = ✓, ✓2 = ✓˙
-3 -2 -1 1 2 3
-1 g 1 g 1
✓˙2 = ✓+ Te = ✓1 + Te
-2 ` m`2 ` m`2
-3
Example 3: Pendulum · =
-sina . 12Te
g 1
✓¨ = sin ✓ + Te (nonlinear equation)
` m`2
For small ✓, use the approximation sin ✓ ⇡ ✓
3 g 1
sinHqL ✓¨ = ✓+ Te
2 ` m`2
q 1
State-space form: ✓1 = ✓, ✓2 = ✓˙
-3 -2 -1 1 2 3
-1 g 1 g 1
✓˙2 = ✓+ Te = ✓1 + Te
-2 ` m`2 ` m`2
-3

✓ ◆ ✓ ◆✓ ◆ ✓ ◆
✓˙1 0 1 ✓1 0
= g + Te
✓˙2 ` 0 ✓ 2
1
m` 2
Linearization
Taylor series expansion:
1
f (x) = f (x0 ) + f 0 (x0 )(x x0 ) + f 00 (x0 )(x x0 )2 + . . .
! 2
⇡ f (x0 ) + f 0 (x0 )(x x0 ) linear approximation around x = x0
Linearization
Taylor series expansion:
1
f (x) = f (x0 ) + f 0 (x0 )(x x0 ) + f 00 (x0 )(x x0 )2 + . . .
2
⇡ f (x0 ) + f 0 (x0 )(x x0 ) linear approximation around x = x0

Control systems are generally nonlinear :

ẋ = f (x, u) nonlinear state-space model

0 1 0 1 0 1
x1 u1 f1
B .. C B .. C B .. C
x=@ . A u=@ . A f =@ . A
xn um fn
Linearization
Taylor series expansion:
1
f (x) = f (x0 ) + f 0 (x0 )(x x0 ) + f 00 (x0 )(x x0 )2 + . . .
2
⇡ f (x0 ) + f 0 (x0 )(x x0 ) linear approximation around x = x0

Control systems are generally nonlinear :

ẋ = f (x, u) nonlinear state-space model

0 1 0 1 0 1
x1 u1 f1
B .. C B .. C B .. C
x=@ . A u=@ . A f =@ . A
xn um fn

Assume x = 0, u = 0 is an equilibrium point: f (0, 0) = 0

This means that, when the system is at rest and no control is
applied, the system does not move.
 system

~
~
↳4
·

Y
- >
- ->

Ax+BU x
·

X
↑ I -
=

->

2x 34+ -- Culin
+

2 x
+

,
CX ,
=

y
-

-
Linearization
Linear approx. around (x, u) = (0, 0) to all components of f :

ẋ1 = f1 (x, u), ..., ẋn = fn (x, u)

Linearization
Linear approx. around (x, u) = (0, 0) to all components of f :

ẋ1 = f1 (x, u), ..., ẋn = fn (x, u)

For each i = 1, . . . , n,
@fi @fi
fi (x, u) = fi (0, 0) + (0, 0)x1 + . . . + (0, 0)xn
| {z } @x1 @xn
=0
@fi @fi
+ (0, 0)u1 + . . . + (0, 0)um
@u1 @um
Linearization /Jacobian Linearization (Using Tay
Linear approx. around (x, u) = (0, 0) to all components of f : s

ẋ1 = f1 (x, u), ..., ẋn = fn (x, u)

expansion)
For each i = 1, . . . , n,
@fi @fi
fi (x, u) = fi (0, 0) + (0, 0)x1 + . . . + (0, 0)xn
| {z } @x1 @xn
=0
@fi @fi
+ (0, 0)u1 + . . . + (0, 0)um
@u1 @um
Linearized state-space model:
@fi @fi
ẋ = Ax + Bu, where Aij =
@xj
, Bik =
@uk
x=0 x=0
u=0 u=0
Linearization
Linear approx. around (x, u) = (0, 0) to all components of f :

ẋ1 = f1 (x, u), ..., ẋn = fn (x, u)

For each i = 1, . . . , n,
@fi @fi
fi (x, u) = fi (0, 0) + (0, 0)x1 + . . . + (0, 0)xn
| {z } @x1 @xn
=0
@fi @fi
+ (0, 0)u1 + . . . + (0, 0)um
@u1 @um
Linearized state-space model:
@fi @fi
ẋ = Ax + Bu, where Aij =
@xj
, Bik =
@uk
x=0 x=0
u=0 u=0

Important: since we have ignored the higher-order terms, this

linear system is only an approximation that holds only for small
deviations from equilibrium.
Example 3: Pendulum, Revisited
Original nonlinear state-space model:

✓˙1 = f1 (✓1 , ✓2 , Te ) = ✓2 — already linear

g 1
✓˙2 = f2 (✓1 , ✓2 , Te ) = sin ✓1 + Te
` m`2
Example 3: Pendulum, Revisited
Original nonlinear state-space model:

✓˙1 = f1 (✓1 , ✓2 , Te ) = ✓2 — already linear

g 1
✓˙2 = f2 (✓1 , ✓2 , Te ) = sin ✓1 + Te
` m`2
Linear approx. of f2 around equilibrium (✓1 , ✓2 , Te ) = (0, 0, 0):
@f2 g @f2 @f2 1
= cos ✓1 =0 =
@✓1 ` @✓2 @Te m`2
@f2 g @f2 @f2 1
= =0 =
@✓1 ` @✓2 @Te m`2
0 0 0
Example 3: Pendulum, Revisited
Original nonlinear state-space model:

✓˙1 = f1 (✓1 , ✓2 , Te ) = ✓2 — already linear

g 1
✓˙2 = f2 (✓1 , ✓2 , Te ) = sin ✓1 + Te
` m`2
Linear approx. of f2 around equilibrium (✓1 , ✓2 , Te ) = (0, 0, 0):
@f2 g @f2 @f2 1
= cos ✓1 =0 =
@✓1 ` @✓2 @Te m`2
@f2 g @f2 @f2 1
= =0 =
@✓1 ` @✓2 @Te m`2
0 0 0

Linearized state-space model of the pendulum:

✓˙1 = ✓2
g 1
✓˙2 = ✓1 + Te valid for small deviations from equ.
` m`2
General Linearization Procedure
I Start from nonlinear state-space model

ẋ = f (x, u)
General Linearization Procedure
I Start from nonlinear state-space model

ẋ = f (x, u)

I Find equilibrium point (x0 , u0 ) such that f (x0 , u0 ) = 0

General Linearization Procedure
I Start from nonlinear state-space model

ẋ = f (x, u)

I Find equilibrium point (x0 , u0 ) such that f (x0 , u0 ) = 0

Note: di↵erent systems may have di↵erent equilibria,
not necessarily (0, 0), so we need to shift variables:

x = x x0 u = u u0
f (x, u) = f (x + x0 , u + u0 ) = f (x, u)
 Linear Transformation

3
General Linearization Procedure f : U ->
Y

I Start from nonlinear state-space model ker (f) =

Im (f) X
=

ẋ = f (x, u)
of
↳thin
...

-
-

Acc be Y =

I Find equilibrium point (x0 , u0 ) such that f (x0 , u0 ) = 0

Note: di↵erent systems may have di↵erent equilibria,
not necessarily (0, 0), so we need to shift variables:

x = x x0 u = u u0
f (x, u) = f (x + x0 , u + u0 ) = f (x, u)

Note that the transformation is invertible:

x = x + x0 , u = u + u0
General Linearization Procedure
General Linearization Procedure
I Pass to shifted variables x = x x0 , u = u u0

ẋ = ẋ (x0 does not depend on t)

= f (x, u)
= f (x, u)

— equivalent to original system

General Linearization Procedure
I Pass to shifted variables x = x x0 , u = u u0

ẋ = ẋ (x0 does not depend on t)

= f (x, u)
= f (x, u)

— equivalent to original system

I The transformed system is in equilibrium at (0, 0):

f (0, 0) = f (x0 , u0 ) = 0
General Linearization Procedure
I Pass to shifted variables x = x x0 , u = u u0

ẋ = ẋ (x0 does not depend on t)

= f (x, u)
= f (x, u)

— equivalent to original system

I The transformed system is in equilibrium at (0, 0):

f (0, 0) = f (x0 , u0 ) = 0

I Now linearize:
@fi @fi
ẋ = Ax + Bu, where Aij =
@xj
, Bik =
@uk
x=x0 x=x0
u=u0 u=u0
General Linearization Procedure
I Why do we require that f (x0 , u0 ) = 0 in equilibrium?
General Linearization Procedure
I Why do we require that f (x0 , u0 ) = 0 in equilibrium?
I This requires some thought. Indeed, we may talk about
a linear approximation of any smooth function f at
any point x0 :

f (x) ⇡ f (x0 )+f 0 (x0 )(x x0 ) — f (x0 ) does not have to be 0

General Linearization Procedure
I Why do we require that f (x0 , u0 ) = 0 in equilibrium?
I This requires some thought. Indeed, we may talk about
a linear approximation of any smooth function f at
any point x0 :

f (x) ⇡ f (x0 )+f 0 (x0 )(x x0 ) — f (x0 ) does not have to be 0

I The key is that we want to approximate a given

nonlinear system ẋ = f (x, u) by a linear system
ẋ = Ax + Bu (may have to shift coordinates:
x 7! x x0 , u 7! u u0 )
General Linearization Procedure
I Why do we require that f (x0 , u0 ) = 0 in equilibrium?
I This requires some thought. Indeed, we may talk about
a linear approximation of any smooth function f at
any point x0 :

f (x) ⇡ f (x0 )+f 0 (x0 )(x x0 ) — f (x0 ) does not have to be 0

I The key is that we want to approximate a given

nonlinear system ẋ = f (x, u) by a linear system
ẋ = Ax + Bu (may have to shift coordinates:
x 7! x x0 , u 7! u u0 )
Any linear system must have an equilibrium point at
(x, u) = (0, 0):

f (x, u) = Ax + Bu f (0, 0) = A0 + B0 = 0.