We define a linear time varying control system to be one of the form {x˙(t)=A(t)x(t)+B(t)u(t)x(t0)=x0where A∈C(J;Mn(R)),B∈C(J;Mn×m(R)).
Existence & Uniqueness
Let J⊂R be any interval, and let t0 be an internal point of J, (i.e. t0∈int(J)). Let A:J→Mn(R) be continuous. Then ∃!x∈C1(J;Rn) that solves {x˙(t)x(t0)=A(t)x(t)=x0 We can then refine this theorem in the case where J is a closed and bounded interval
Let a,b∈R with a<b and let A:[a,b]→Mn(R) be continuous. Suppose t0∈(a,b), then ∃!x∈C1([a,b];Rn) that solves {x˙(t)x(t0)=A(t)x(t)=x0 In class we first proved 2 then used 2 to easily prove 1.
In the course of proving the above, we introduced a limit matrix ΦA(t,t0) where ΦA(t,t0)=k→∞limXk(t)where we defined {Xk(t)}k∈N as a sequence of functions on Mn(R) (i.e. {Xk(t)}k∈N⊂Mn(R)) that converge to a solutionX(t):[a,b]→Mn(R)
Let J⊆R be a non-empty interval. Let t0 be an internal point of J and A:J→Mn(R). Then {X˙(t)X(t0)=A(t)X(t)=Inhas a uniqueC1 solution which is given by our transition matrixΦA(t,t0).
We then went on to prove the same for the controlled variant
Let J⊂R be a non-empty interval. There is a unique solution to the following LTVC system{x˙(t)x(t0)=A(t)x(t)+B(t)u(t)=x0given by x(t)=ΦA(t,t0)x0+t0∫tΦA(t,τ)B(τ)u(τ)dτwhere ΦA(t,t0) is the Transition Matrix.
Controllability
Consider a LTVC system{x˙(t)=A(t)x(t)+B(t)u(t)x(t0)=x0Let t0,t1∈J, t0<t1 and let x0,x1∈Rn. Then, ∃ some control u∈C([t0,t1];Rm) which “steers” the system from x(t0)=x0 to x(t1)=x1if and only ifΦA(t0,t1)x1−x0∈Image(W(t0,t1))holds, where W(t0,t1)=t0∫t1ΦA(t0,τ)B(τ)BT(τ)ΦAT(t0,τ)dτis our controllability gramian. Therefore, the control input u(t) is defined as u(t)=BT(t)ΦAT(t0,t)η where η∈Rn s.t. ΦA(t0,t1)x1−x0=W(t0,t1)η gives the desired transfer.
Observability
Consider the LTVC System introduced in the Observability Problem: x˙(t)=A(t)x(t)+B(t)u(t)y(t)=C(t)x(t)+D(t)u(t)The initial state can be uniquely determined (i.e. our system is Observable) if Ker(M(t0,t1))={0}
Realization
Suppose (A,B,C) is a LTVC SystemRealization of T:J×J→Mp×n(R). Then (A~,B~,C~) with A~(t)B~(t)C~(t)=P˙(t)P−1(t)+P(t)A(t)P−1(t)=P(t)B(t)=C(t)P−1(t)where P:J→Mn×n(R) is C1 and non singular is also a Realization.