Lecture 9: Controllability Flashcards
Controllability
transfers state to origin
x(tf) = 0; x(tf) = -Iota(tf,t0)*x0
Reachability
set of all states x1 reachable at t1
x0 = int(Iota(tf,Tau)B(Tau)u(Tau)dTau
Relationship Between Reachability/Controllability
Completely reachable -> completely controllable
Conpletely controllable -> completely reachable if transitions matrix is non singular (CT systems always true)
A-invariant
if x -> controllable subspace then Ax -> controllable subspace
Reachability in Controllable Subspace
Any element of Controllable subspace can reach any other element within the controllable subspace
Controllability Canonical Form
T = (T1,T2) T1 = span of q-dimensional controllable subspace T2 = form basis with T1
Controllability Transformation Matrix applied to System
A' = [A'11 A'12; 0 A'22] = inv(T)*A*T B' = [B'1; 0] = inv(T)*B
Properties of Controllability Transformation Matrix
- (A’11,B’1) is completely controllable
- A’11 controllable eigenvalues
- A’22 uncontrollable eigenvalues
- x’ = inv(T) * x -> controllability canonical form
Methods of determining Controllable for LTV
- If grammian yields non-singular matrix
2. M-test must have rank = dim(A)
Controllaibility for DT Linear systems
- If Iota has full rank then reachability matrix must have full rank
- If Iota has less than full rank then Rc rank must be < n
DT Reachability Matrix
- Rc = (B(k1 - 1), Iota(k1,k1-1)B(k1-2),…, Iota(k1, k0 + 1)B(k0))
- Rc = (B, AB, A^2B, … , A^(n-1)*B) for time invariant
M-Test
M0(t) = B(t) Mj(t) = -A(T)*M(j-1)(t) + M'(j-1)(t), j = 1, ... , n-1