CHAPTER 1.

REVIEW OF CLASSICAL CONTROL THEORY

1.1 Classical control theory

: deals with linear, Constant coefficient SISO systems (= time invariant)
- Linear system : - the equations of the model are linear
- the principle of superposition is applicable.
- nonlinearities

saturation nonlinearity dead-zone nonlinearity square-law nonlinearity

In linear system, analytic and graphical techniques is easy in control design and
analysis.

- Classical(conventional) Control
: SISO(single-input and single-output) system Control,
main control factors: input, output, error

- Modern Control
: MIMO(Multi-input and Multi-output)system Control,
main control factor: state variable
( → applicable to linear or nonlinear time invariant or
time varying systems)

1.2 System representation

Linear, stationary system is simplified by the use of transform techniques
For continues-time system, Laplace transform used
For discrete-time system, Z transform used
Linear system
: It obeys superposition principle

Linear
u1(t) → y1(t)
System
Linear
a1u1(t)+a2u2(t)→
(t)→ →a1y1(t)+a2y2(t)
System
Linear
u2(t) → → y2(t)
System

superposition

ex) F = kx linear system

F = kx 2
nonlinear system
F = kx + b nonlinear system

Linear time-invariant differential system

- System is linear
- Coefficient is time-invariant

ex)
d 2x dx
Linear: a + b + cx = 0
dt dt
Nonlinear:
2
d 2x  dx 
a + b  + cx = 0
dt  dt 
d 2 x dx
+ + cx 2 = 0
dt dt
 • Laplace Transformation
- Laplace Transformation : mathematical tool used for the solution of
ordinary linear differential equations. Converting many functions into algebraic
functions of a complex variable

Definition
∞
Given f(t) such that ∫ f (t ) ⋅ e −σ t dt < ∞ ,
0

∞
Laplace Transform : F(s)= ∫ f (t )e − st dt or, F ( s ) = L[ f (t )]
0

(where, s = σ + jw is complex variable)

ex) f (t ) = e − at t≥0 ,where a=constant

∞ ∞
− at − st
∞
−( s + a ) t e −( s + a ) t
F ( s ) = ∫ e e dt = ∫ e dt = −
0
0 ( s + a) 0

1
= --- ①
s+a

, where equation ① is satisfied when

∞
∞
− at −σt e −(σ + a )t
∫0
e e dt = −
(σ + a ) 0
< ∞

Therefore, for condition σ > −a , Laplace transform exists.

- Inverse Laplace Transform

Laplace Transform F(s) → f(t) : inverse Laplace transform
f (t ) = L−1[ F ( s )]
1 c + j∞
=
2πj ∫c − j ∞
F ( s )e st ds

, where C is real constant and is chosen larger than the real parts of all singular
points of F(s)
Important Theorems of the Laplace Transform
1. Multiplication by a constant
L[ Kf (t )] = KF ( s )

2. Sum and difference

L[ f1 (t ) ± f 2 (t )] = F1 ( s) ± F2 ( s)
3. Differentiation
 df (t ) 
L  = SF ( s ) − f (0)
 dt 
for higher-order derivatives,

 d n f (t ) 
 = S F ( s) − S f (0) − S f ′(0) … − Sf
n −1 n− 2 ( n −2 )
L n
n
(0) − f ( n −1) (0)
 dt 

Laplace Transform Pairs

f(t) F(s)

unit impulse δ (t ) 1
1
unit step u(t)
s
- 1
t
s2
1
e − at
s+a

Application of Laplace Transform

d 2 x(t ) dx(t ) 1(t > 0)
ex) + 3⋅ + 2 x(t ) = 5u s (t ) ,where u s (t ) =  ; unit step function
0(t < 0)
2
dt dt
dx(t )
Initial conditions : x(o) = −1, x′(0) = =2
dt t =0
Solve the equation using Laplace transform
sol) We first take the Laplace transform :
5
{S 2 X ( s ) − Sx (0) − x′(0)} + 3{S ⋅ X ( s ) − x (0)} + 2 X ( s ) =
s
 Substiting the values of x(0) and x′(0) into above Eq.

5
S 2 X ( s ) + S − 2 + 3SX ( s ) + 3 + 2 X ( s ) =
s
5
→ S 2 X ( s ) + 3SX ( s ) + 2 X ( s ) + S + 1 =
s
5
→ X ( s )( S 2 + 3S + 2) + S + 1 =
s
− S2 − S +5
→ X ( s )( S 2 + 3S + 2) =
S
−S −S +5
2
− S2 − S +5
→ X ( s) = =
S ( S 2 + 3S + 2) S ( S + 1)( S + 2)
A B C A( S + 1)( S + 2) + BS ( S + 2) + CS ( S + 1)
= + + =
S S +1 S + 2 S ( S + 1)( S + 2)
A( S 2 + 3S + 2) + B( S 2 + S ) + C ( S 2 + S )
=
S ( S + 1)( S + 2)

 A + B + C = −1
 5
 3 A + 2 B + C = −1 ∴ A=
 2A = 5 2


 7
B +C = −
 2 3
∴ B = −5, C=
 17 2
 2B + C = −
 2
By partial – fraction expansion ,
5 5 3
X ( s) = − +
2 S S + 1 2( S + 2)
Taking the inverse Laplace transform
5 3
x(t ) = − 5e −t + e − 2 t (t ≥ 0)
2 2

↑ ↑
Steaty-
teaty-state transient period
period

● Transfer function of Linear System

: ratio of Laplace transform between input and output variable
: may be defined only for a linear, stationary (constant parameter) system.
 For above spring-mass-damper system
ΣF = Mɺyɺ
r (t ) − Cyɺ (t ) − ky (t ) = Mɺyɺ(t )
Mɺyɺ(t ) + Cyɺ (t ) + K ⋅ y (t ) = r (t )
Taking Laplace transform,
MS 2Y ( s ) + CSY ( s ) + KY ( s ) = R( s )

(at zero initial condition i.e., y (o) = o, y′(o) = 0 )

Then, transfer function is noted
Y (s) 1
G (s) = =
R( s ) MS + CS + K
2

1.3 Performance and Analysis in classical Control Theory

Consider feedback control system,
R(s) + E(s) C(s)
Gc(s) Gp(s)
-
F(s)

H(s)

R(s) + E(s) C(s)

G(s)
-
F(s)

H(s)

( G(s)=Gc(s)*Gp(s) : forward transfer function, H(s) : feedback transfer

function )
At the summing junction, R - HC = E
Between E and C, GE = C
Elimination of E gives :
C
R - HC =
G

(GH + 1)
→ R= C
G
G
∴C = R
1 + GH

C 1
E= = R
G 1 + GH

G ( s) R( s)
Now, C ( s) = 
→ analytical Solution for C(t)
1 + G( s) H ( s)
Inverse Laplace transform

※ Without analytical solutions, possible to evaluate performance

: analyzing the behavior of C(t)

Common test inputs : Step functions, ramps, and sinusoids.

Checking:
stability
steady-state error
transient response
frequency response
sensitivity due to disturbances and vibrations

1.4 Stability analysis

Stability : Output C(t) must not grow without bound due to a bounded input,
or bounded disturbance.
 In the below figure, we can say the ball has not stability. For a small
disturbance, the ball will roll down without bound.

There are two concepts in stability:

-absolute stability (yes or no in decision)
-relative stability (how much stable or not in decision)

a) Routh - Hurwitz Criterion (for absolute stability)

C ( s ) bo S n + b1 ⋅ S n −1 + … + bn −1 ⋅ S + bn
For closed loop transfer function: =
R( s ) a o S n + a1 ⋅ S n −1 + … + a n −1 ⋅ S + a n

Characteristic equation 
→ ao S n + a1 ⋅ S n −1 + … + an −1 ⋅ S + an =0

The process of Routh - Hurwitz Criterion is as follows:

1. Coefficients in two rows :

a0 a2 a4 a6 …
a1 a3 a5 a7 …

2. Form array (for example, sixth-order system is considered)

S6 ao a2 a4 a6

S5 a1 a3 a5 0
S 4
a1a2 − ao a3 a1a4 − ao a5 a1a6 − ao ⋅ 0
=A =B = a6 0
a1 a1 a1
S3 A ⋅ a3 − a1 ⋅ B A ⋅ a5 − a1a6 A ⋅ 0 − a1 ⋅ 0
=C =D =0 0
A A A
BC − AD C ⋅ a6 − A ⋅ 0 C ⋅0 − A⋅0
S2 =E = a6 =0 0
C C C
ED − C ⋅ a6
S1 =F 0 0 0
E
F ⋅ a6 − E ⋅ 0
S0 = a6 0 0 0
F

Above table is called Routh table

3. Investigate the signs of numbers in the first column of the tabulation

<Conclusion>
i) If all the elements are of the same sign
→ all roots are in the left half of the S plane
(which means that the system is stable)
ii) If change of signs in the elements
→ the number of sign changes = number of roots with positive real
parts. (which means that the system is unstable)

Ex) Consider the equation

(S-2)(S+1)(S-3)= S 3 − 4S 2 + S + 6 = 0
↑ ↑ ↑ ↑
ao a1 a2 a3
sol) Routh tabulation :

S3 ao =1 a2 =1

S2 a1 = −4 a3 = 6
( − 4) × 1 − 1 × 6
S1 = 2 .5 0
−4
2 .5 × 6 − ( − 4 ) × 0
S0 =6 0
2 .5

∴ two sign changes ⇒ We can say that two roots located in the right half of the
S-plane by Routh tabulation. Therefore, the system is unstable.

In fact we know that the roots are 2, -1, and 3 from equation (S-2)(S+1)(S-3)=0.

-1 2 3

(Note) Routh-Hurwitz gives only an absolute answer on stability

b) Root Locus Method (for relative stability) → more versatile for analysis and
design problems

R(s) C(s)
+ G(s)
-

H(s)

- closed-loop transfer function :

C (s) G (s)
=
R( s) 1 + G ( s) H (s)

- characteristic equation :
1 + G(s)H(s) = 0 or G(S)H(S) =-1

Since G(S)H(S) is a complex quantity , it can be split into angle and magnitude
angle condition :

∠G ( s ) H ( s ) = ±180 °( 2k + 1) , ( k = 0,1,2 ⋯) ①
---①

magnitude condition :

G( s) H (s) = 1 ②
---②

∴ The values of S which satisfy ① and ② are the roots of the characteristic
equation. (or closed loop poles)

* Root-locus plots of second-order systems

R(s) C(s)
+ K
- s ( s + 1)
- loop transfer function
k
G (s) H ( s) =
S ( S + 1)
- closed-loop transfer function
k
C (s) k G (s) S ( S + 1) k
= 2 (← = = 2 )
R( s) S + S + k 1 + G( s) H ( s) 1 + k S +S +k
S ( S + 1)
- characteristic equation : S2 + S + k = 0

Now, we want to find the locus of roots of characteristic equation as K is varied

from zero to infinity.

- roots of the characteristic equation :

1 1 1 1
S1 = − + 1 − 4k , S2 = − − 1 − 4k
2 2 2 2
1
if k≤ , roots = real ( overdamped system)
4
1
if k> , roots = complex (underdamped system)
4

1 1 1
ex) Let’s check the point p = − + i ( k = ) on the locus:
2 2 2
k 0 .5
G (s) = =
S ( S + 1) ( −0.5 + 0.5i )(0.5 + 0.5i )
0 .5
= = −1
− 0.25 − 0.25
 k 
∠ = 180°
S ( S + 1) s = p 
∴ 
k
=1 
S ( S + 1) s = p 


∴ O⋅K (the point p is a root of the characteristic equation).

k
Ex) ∠ = −∠S − ∠S + 1 = ±180°(2n + 1), (n = 0,1,2, ⋯)
S ( S + 1)
1
If we choose p = − + 2 j, what value is k?
2

k 17
=1 ∴k =
1 1 4
(− + 2 j )( + 2 j )
2 2
(If we choose k=17/4, then the root of the characteristic equation can be p )

(Note) Root Locus drawing using Matlab

G(s)
R(s)
+
K ( s + 1) C(s)

s ( s + 2)
-

1
s+3
H(s)

i) consider closed-loop control system

C ( s) k ( S + 1)( S + 3)
T (s) = =
R ( s ) S ( S + 2)( S + 3) + k ( S + 1)

- Characteristic equation :
S(S+2)(S+3) + K(S+1) =0
P(s)
ii) rewrite rhe form as 1 + k =0
q(s)

s +1
In this case, 1 + k =0
s ( s + 2)( s + 3)
 iii) matlab program

〉p=[1 1];

〉q=[1 5 6 0];

〉rlocus(p,q)

Root Locus
8

2
Imaginary Axis

-2

-4

-6

-8
-3 -2.5 -2 -1.5 -1 -0.5 0
Real Axis

[Home Work1]
d 2 x(t ) dx(t ) 1(t > 0)
[1] + 5 ⋅ + 4 x (t ) = u (t ) ,where u (t ) =  ; unit step function
0(t < 0)
s s
dt 2 dt
dx(t )
Initical conditions : x(o) = −1, x′(0) = =2
dt t =0
Solve the equation using Laplace transform

[2] Consider the below feedback system,

R(s) C(s)
+ K ( s + 2)
- s 2 + 2s + 3

i) Draw root-locus plot of the control system by using MATLAB

ii) What is roots of characteristic equation for k=0.5, 1, respectively?
ⅲ) Check the step response and compare the performances for k=0.5, 1,
respectively.