Control
Control
V. Sankaranarayanan
Outline
1 Controllability
Controllability
ẋ = Ax + Bu (1)
y = Cx + Du (2)
where
x ∈ Rn ; u ∈ Rm ; y ∈ Rp
A ∈ Rn×n B ∈ Rn×m
C ∈ Rp×n D ∈ Rp×m
Definition
A linear system, described above by state space equations (1) and is said to be
controllable, if for any initial state x(0) = x0 and any final state x(T ) = xf , there
exists an unconstrained control input u(t),0 ≤ t ≤ T that transfers the system from
x0 to xf in a finite time ’T ’. Otherwise the system is said to be uncontrollable.
Controllability
Definition
A state x0 is said to be Controllable, if there exists a finite time interval [0, T ]
and an input u(t), t ∈ [0, T ] such that x(T ) = 0
If all sates are controllable, then the system is said to be Completely
Controllable
Proof
Rt
x(t) = eAt x(0) + 0 eA(t−τ ) Bu(τ )dτ
Since the
R controllability depends only on input state variables.
x(t) = 0t eA(t−τ ) Bu(τ )dτ
x(t) = 0t eAt Bu(t − τ )dτ
R
(At)2 (At)3
eAt = I + At + 2! + 3!
+ ....
eAt = n−1 i
P
i=0 A αi (t)
Controllability
Controllability
Basic Pre-requisites
Before we prove the condition for controllability let us see some basic
prerequisites.
Consider, the matrix
λ t
teλ1 t t2 eλ1 t
e 1 0 0
0
eλ1 t teλ1 t 0 0
−1
eAt = Q 0 0 eλ1 t 0 0 Q
0 0 0 eλ1 t 0
0 0 0 0 eλ2 t
We can see that every entry of eAt is a linear combination of terms
[eλ1 t , teλ1 t , t2 eλ1 t , eλ2 t ].
These values depend upon eigenvalues and their indices. Here,n̄1 − 1 = 2,
where n̄1 is the index of the eigenvalue λ1
In general, if A has an eigenvalue with index n̄1 , then every entry of eAt is a
linear combination of terms eλ1 t , teλ1 t , t2 eλ1 t , · · · , tn̄1 −1 eλ1 t
Every such term can be infinitely differentiable and can be expanded in a
Taylor series at every ’t’ and is called Analytic.
Controllability
Controllability
Let us claim that for any x(0) = x0 and any state x(t1 ) = x1 , the input
T
u(t) = −B T eA (t1 −t) W −1
c (t1 )[e
At x − x ] will transfer the state x to x in
0 1 0 1
the interval t1
Substituting u(t) in the above equation we get ,
Z t T
(t1 −t)
x(t1 ) = eAt x(0) − eA(t−τ ) BB T eA W −1
c (t1 )[e
At
x0 − x1 ]dτ
0
x(1) = eAt x(0) − W c W −1
c (t1 )[e
At
x0 − x1 ]
Controllability
Let,C = eAτ B
We know that eAτ B are analytic, i.e, every such term can be infinitely
differentiable and can be expanded in a taylor series at every ’t’.
Therefore, W c is non singular if and only if there exists no n × 1 non zero
vector q such that q T C = 0
Therefore, C has full row rank,i.e, rank of C = n
Controllability
ẋ1 = −2x2 + u
ẋ2 = x1 − 3x2 + u
y = x1
ẋ1 0 −2 x1 1
= + u
ẋ2 1 −3 x2 1
x1
y = 1 0
x2
Controllability
C = B AB 2×2
1 −2
∴C =
1 −2
Controllability
Controllability
ż1 −2 0 z1 1
= + u
ż2 0 −1 z2 0
1 −2
Controllability matrix, C =
0 0
Rank of the matrix C = 1, has a row full of zeroes,is less than 2 .
Therefore, the system is not Controllable.
Controllability
Controllability
ż1 0 1 z1 1/3
= + u
ż2 −2 −3 z2 −2/3
1/3 −2/3
Controllability matrix, C =
−2/3 4/3
Rank of the matrix C = 1, is less than 2 .
Therefore, the system is not Controllable.
Controllability
ẋ1 = 3x2
ẋ2 = −2x1 + 5x2 + u
y = x1
ẋ1 0 3 x1 0
= + u
ẋ2 −2 5 x2 1
x1
y = 1 0
x2
Controllability
C = B AB 2×2
0 3
∴C =
1 5
Controllability
Controllability
ż1 2 0 z1 3
= + u
ż2 0 3 z2 −3
3 6
Controllability matrix, C =
−3 −9
Rank of the matrix C = 2.
Therefore, the system is Controllable.
Controllability
Controllability
ż1 0 1 z1 0
= + u
ż2 −6 5 z2 1
0 1
Controllability matrix, C =
1 5
Rank of the matrix C = 2 .
Therefore, the system is Controllable.