EE611 Deterministic Systems Realizations, State-Transition Matrices Kevin D. Donohue Electrical and Computer Engineering University of Kentucky Realizations Every LTI system has an input-output description of the form: s u s y s= G If system is also lumped, state-space descriptions also exist: ẋ t=A x tB ut y t =C x tD ut G s which are referred to as realization of the transfer matrix . G s is realizable if ∃ a finite-dimensional state equation {A,B,C,D} ∋ G s = DC s I−A −1 B Transfer Matrix Realizations A transfer matrix G s is realizable iff G s is proper rational matrix. Consider an element of G s that is not proper (M>0): M M −1 g p ,q s =d 0 s d 1 s ...d M −1 sd M b1 sn−1b2 sn−2...bn−1 sbn sn a1 s n−1 a2 sn−2...a n−1 sa n How would the realization equations below have to change to accommodate the not proper G s ? ẋ t=A x tB ut y t =C x tD ut Example Find the state equations for the circuit using 2 different state definitions. Show the transfer matrix is the same. Assume u(t) input and y(t) output. u(t) R L C + y(t) - 1 LC g s= R 1 2 s s L LC Realize a Strictly Proper TF For a proper TF, long division can be applied to decompose it into a constant term (for the d scalar) and a strictly proper TF expressed below: g sp s= 1 sn−1 2 sn−2... n−1 sn sn 1 sn−1 2 s n−2...n−1 s n The op-amp circuit for this TF can be realized as: β1 u(t) β2 β3 1/S −α1 1/S −α2 −α3 1/S −αn ..... 1/S βn y(t) Realize a Strictly Proper TF The state equations for the strictly proper TF: g sp s= 1 sn−1 2 sn−2... n−1 sn sn 1 sn−1 2 s n−2...n−1 s n are written from: x 1= ẋ 2 x 2 = ẋ 3 ... x n−1= ẋ n ẋ 1=u−1 x1 −2 x 2 ...−n−1 x n−1 − n x n y=1 x 12 x 2 ...n−1 x n−1 n x n [ −1 − 2 0 ẋ= 1 ⋮ ⋮ 0 0 ] [] ... −n−1 −n 1 ... 0 0 x 0 u 0 ⋮ ⋮ ⋮ 0 ... 1 0 y= [ 1 2 ... n−1 n ] x Transfer Matrix Realizations In general for a proper q x p transfer matrix G s its realization can be expressed as follows. Find a common denominator d(s) for all element in G s and divide through by the denominator (if necessary) to separate each rational polynomial into a constant and strictly proper rational polynomial s= G∞ s G G sp Then expand the strictly proper part into: 1 r−1 r−2 G sp s= N s N s ... Nr−1 sNr ] [ 1 2 d s where r is the order of d(s). Transfer Matrix Realizations A realization of G scan therefore be written as: [ ] [] −1 I p − 2 I p ... − r−1 I p − r I p Ip Ip 0 ... 0 0 ẋ= x 0 u ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ 0 0 0 ... Ip 0 y=[ N1 N 2 ... Nr−1 Nr ] xG∞u Solution of Linear Time-Varying State Equation Given a linear-time varying state-space equation: ẋ t=A t x tB t ut y t =Ct x tD t ut the solution can be written as: t x t= t ,t 0 x t 0 ∫ t , B ud t0 t y t =Ct t , t 0 xt 0 Ct ∫ t , B ud Dtut t0 where t , is called the state-transition matrix State Transition Matrix A fundamental matrix for the homogeneous equation ẋ t=A t x t is a matrix X(t) ∋ its columns are unique solutions of: Ẋ t =A t X t and X(t) is nonsingular for all t. Then for any fundamental matrix of ẋ t=A t x t the state transition matrix is given by: t , t 0 =X t X−1 t 0 and is the unique solution of: ̇ t , t 0 =A tt , t 0 for initial condition t 0, t 0 =I Examples Find the state transition matrices for to the following: [ ] ẋ t= [ 0 0 x t t 0 ] 1 exp−t ẋ t= xt 0 −3 Special Case Solution For the case when A t t ∫ A d t0 = t ∫ A d t0 A t Then the solution for the state transition matrix becomes: ̇ t 0, t =exp t ∞ ∫ A d =∑ t0 k =0 1 k! t ∫ A d t0 k Discrete Time Case The state transition matrix can be computed through recursion for the discrete time case. Given k 1, k 0 =A [k ] k , k 0 Substitute repeatedly into k , k 0 =A [k −1]k −1, k 0 k , k 0 =A [k −1] A [k −2] k −2, k 0 k , k 0 =A [k −1] A [k −2]..... A [k 0 ] k 0, k 0 Then the solution for the state transition matrix becomes: k , k 0 =A [k −1] A [k −2]..... A [k 0 ] Useful Matrix Relationships −1 d A t d A t −1 −1 =−A t A t dt dt ∣A B∣=∣A∣∣B∣ −1 A B =B−1 A−1 T A B =BT AT tr A B =tr B A Homework U6.1 Find the op-amp circuit and state-space realization for the proper rational TF given below: 2 s33 s g s= 3 s 2 s2 5 s25