Control Engineering

Ramkrishna Pasumarthy

Department of Electrical Engineering,

Indian Institute of Technology Madras
 Module 12
Lecture 1

Controllability Matrix and Controllable Systems

Control Engineering Module 12 Lecture 1 Ramkrishna P. 1/10

Controllability Matrix
Controllability Matrix

Consider the continuous time LTI system

ẋ = Ax + Bu, x ∈ Rn , u ∈ Rk (1)

For the system in (1), the controllability Gramians is given, respectively, by

∫ t1 ′ ∫ t1 −t0 ′
WC (t0 , t1 ) := t0
eA(t0 −τ ) BB′ eA (t0 −τ ) dτ = 0
e−At BB′ e−A t dt

The controllability matrix of the system in (1) is defined as

C := [B AB A2 B · · · An−1 B]n×(kn) (2)

Control Engineering Module 12 Lecture 1 Ramkrishna P. 2/10

Example: RC Circuit

R1 R2
u

+ +
x1 C1 x2 C2
− −

Figure 1: Parallel RC cirucit

Consider the parallel RC circuit shown in figure 1. The controllability matrix for the
electrical circuit is given by
[ 1 ]
R1 C1 − R21C2
C = [B AB] = 1
1 1

R2 C2 − R21C2
2 2

Control Engineering Module 12 Lecture 1 Ramkrishna P. 3/10

Example: RC Circuit
1 1
▶ When two branches have the same time constant, i.e. R1 C1 = R2 C2 =: ω, we have
[ ]
ω −ω 2
C=
ω −ω 2
{ [ ] }
and therefore { } 1
C [t0 , t1 ] = Im C = α : α ∈ R , ∀ t1 > t0 ≥ 0.
1

▶ However, when the time constants are different, i.e. R11C1 = R21C2 ,
( )
1 1 1
det (C) = − ̸= 0
R1 C1 R2 C2 R1 C1 R2 C2
which means that C is nonsingular and therefore
{ }
R[t0 , t1 ] = C [t0 , t1 ] = Im C = R2

Control Engineering Module 12 Lecture 1 Ramkrishna P. 4/10

ControllabLe Systems
Controllable Systems

Consider the following continuous time LTV system

ẋ = A(t)x + B(t)u, x ∈ Rn , u ∈ Rk (3)

Definition 7.2.1
Given two times t1 > t0 ≥ 0, the system in (3), or simply the pair (A(.), B(.)), is (completely
state- ) controllable on [t0 , t1 ] if C [t0 , t1 ] = Rn i.e. if every state can be transferred to the
origin.

Control Engineering Module 12 Lecture 1 Ramkrishna P. 5/10

Controllable Systems

Consider the LTI system given in (1). From Theorem 7.2.1 we have
{ }
Im C = C [t0 , t1 ]

Theorem 7.2.2

The LTI system (1) is controllable if and only if

{ }
rank C = n

Control Engineering Module 12 Lecture 1 Ramkrishna P. 6/10

Example: Pendulum on a Cart

Consider the dynamics of the cart-pendulum system discussed in Lecture 1. The

dynamics of the cart-pendulum system is
     
x ẋ 0
d     
0

θ  
θ̇  
 

u
ż =   =  Jt mlsθ θ̇ −kJt ẋ+cmlcθ θ̇+m l gcθ sθ  + 
2 2 2
Jt 
dt ẋ  Mt Jt −m2 l2 c2θ   Mt Jt −m2 l2 c2θ 
2 2 2 mlcθ
θ̇ kmlcθ ẋ−m l cθ sθ θ̇ −cMt θ̇−Mt mglsθ − Mt Jt −m
2 2 2 2 l2 c2
Mt Jt −m l cθ θ

The equilibrium points of the system are [x 0 0 0]T i.e. the pendulum with upright
position and [x π 0 0]T , the pendulum with down position, ∀ x ∈ R.

Control Engineering Module 12 Lecture 1 Ramkrishna P. 7/10

Example: Pendulum on a Cart

Assume that the system stays within a small neighbourhood with the pendulum in the
upright position. Then
sin θ ≈ θ, cos θ ≈ 1, θ̇2 ≈ 0

∴ The linearized dynamics leads to a set of linear state space equation

   
0 0 1 0 0
0 0 0 1   0 
   
ż =  m2 l 2 g kJt cml  z +  Jt u
0 Mt Jt −m2 l2 − Mt Jt −m 2 l2 Mt Jt −m l2 2   Mt Jt −m l 
2 2
ml
0 − MtMJtt−m
mgl
2 l2
kml
Mt Jt −m2 l2 − Mt JtcM t
−m2 l2
− Mt Jt −m 2 l2
| {z } | {z }
A B

Control Engineering Module 12 Lecture 1 Ramkrishna P. 8/10

Example: Pendulum on a Cart

Q. Can the system be moved from one stationary point to another by appropriate
application of forces through the wheels?
The controllability matrix of the linearized dynamics is given by C = [B AB A2 B A3 B]1

g2 J2t l4 m4 M2t g2 l8 m8 2g2 Jl6 m6 Mt

∴ det (C) = − 6
− 6
+ 6
(Jt Mt − l2 m2 ) (Jt Mt − l2 m2 ) (Jt Mt − l2 m2 )
g2 l4 m4
=− 4
(Jt Mt − m2 l2 )

▶ For finite value of det(C), Jt Mt ̸= m2 l2 which implies that det(C) ̸= 0.

▶ ∴ rank (C) = 4 i.e. the system is controllable.
1 The complete controllability matrix is shown in the appendix

Control Engineering Module 12 Lecture 1 Ramkrishna P. 9/10

Overview

Summary: Module 12 Lecture 1 Contents: Module 12 Lecture 2

▶ Controllability Matrix ▶ Controllability Tests
▶ Controllable Systems

Control Engineering Module 12 Lecture 1 Ramkrishna P. 10/10

Pendulum on a Cart: Controllability Matrix

C = [B AB A2 B A3 B], where

   cl2 m2 +J2 k

  Jt − (l2 m2 −J Mt )2
0 Jt Mt −l2 m2  t t

   lm   lm(cMt +Jt k) 
 0   l2 m2 −Jt Mt   2 m2 −J M )2 
B=  , AB =  cl2 m2 +J2 k  , A2 B = 
(l t t

Jt
 Mt Jt −m2 l2  − t   − c2 l2 m2 Mt +Jt l2 m2 (2ck−glmMt )+gl5 m5 +J3t k2

 (l2 m2 −Jt Mt )2   (l2 m2 −Jt Mt )3 
ml lm(cMt +Jt k)
− Mt Jt −m 2 l2
2 2 2 2 3 3
lm(c Mt +Jt Mt (ck−glmMt )+ckl m +gl m Mt +Jt k ) 2 2
(l2 m2 −Jt Mt )2
(l2 m2 −Jt Mt )3

 c2 l2 m2 M +J l2 m2 (2ck−glmM )+gl5 m5 +J3 k2


− t t
(l2 m2 −Jt Mt )3
t t

 
 lm(c Mt +Jt Mt (ck−glmMt )+ckl m +gl m Mt +J2t k2 )
2 2 2 2 3 3

 (l2 m2 −Jt Mt )3 
A B =  2Jt l2 m2 (c2 kMt −cglmM2 +gkl3 m3 )+cl2 m2 (c2 M2 +ckl2 m2 +2gl3 m3 Mt )+J2 kl2 m2 (3ck−2glmMt )+J4 k3 
3 

− t t t t

 (l 2 m2 −J M )4
t t 
lm(c3 M3t +2c2 kl2 m2 Mt +J2t kMt (ck−glmMt )+cJt (ckM2t −2glmM3t +2k2 l2 m2 )+2cgl3 m3 M2t +gkl5 m5 +J3t k3 )
(l2 m2 −Jt Mt )4

Control Engineering Module 12 Lecture 1 Ramkrishna P.