4-1 Chapter 4 State-Space Solutions and Realizations Discuss solutions of state-space and study how to transform transfer functions into state equations 長庚大學電機系 4-2 Outline Solution of linear time-invariant (LTI) state equations Equivalent state equations Realizations Solution of linear time-varying (LTV) equations Equivalent time-varying equations Time-varying realizations 長庚大學電機系 4-3 Solution of LTI State Equations Consider the LTI state-space equation x(t ) Ax(t ) Bu(t ) y(t ) Cx(t ) D(t ) Solution in s-domain (by Laplace transform) x(t ) Ax(t ) Bu(t ) sxˆ ( s) x(0) Axˆ ( s) Buˆ ( s(Take ) Laplace transform) xˆ ( s ) ( sI A) 1[ x(0) Buˆ ( s)] yˆ ( s ) C ( sI A) 1[ x(0) Buˆ ( s)] Duˆ ( s) C ( sI A) 1 x(0) [C ( sI A) 1 B D]uˆ ( s) 長庚大學電機系 4-4 Solution in t-domain: x Ax Bu e At x e At ( Ax Bu ) (Multiply integrating factor) d At (e x) e At Bu dt e At t x(t ) x(0) e A Bu ( )d 0 t x(t ) e x(0) e A(t ) Bu ( )d At 0 t y (t ) C[e x(0) e A(t ) Bu ( )d ] Du (t ) At 0 長庚大學電機系 Computation of e At 4-5 At How to compute e ? At By Laplace transform: e L1[(sI A)1 ] Function of square matrix: k ( At ) At Power series method: e k! k 0 Jordan form: A QJQ1 e At QeJt Q1 長庚大學電機系 4-6 0 1 Example: Compute e , A 1 2 At By Laplace transform method: ( s 2) 1 2 s 1 ( s 1) 1 ( sI A) 1 1 s 2 ( s 1) 2 t (1 t ) e e At L1[( sI A) 1 ] t te 1 ( s 1) 2 s ( s 1) 2 te t t (1 t )e 長庚大學電機系 4-7 By Jordan form method: 1 1 1 1 1 A QJQ , Q , J 1 0 0 1 t t 1 e te At Jt 1 e Qe Q Q Q t 0 e (1 t )e t te t t t (1 t )e te 長庚大學電機系 4-8 By function of square matrix method: Let f ( ) et and h( ) 0 1 ( A) {1, 1} f (1) e t h(1) 0 1 f '(1) te t h '(1) 1 0 (1 t )e t , 1 te t e At f ( A) h( A) 0 I 1 A (1 t )e t t te te t t (1 t )e 長庚大學電機系 4-9 Example: From the previous example, the solution of 0 1 0 x x u 1 2 1 is t x(t ) e x(0) e A(t ) Bu ( )d At 0 (1 t )e t t te t ( t ) t ( t ) e u ( ) d te 0 x (0) t ( t ) (1 t )e t [1 (t )]e u ( )d 0 長庚大學電機系 Stability of Zero-Input Response 4-10 Let A QJQ1. Then Zero-input response xzi (t ) e At x(0) QeJt Q1x(0) k 1 Since t t t t e Jk ( )t e te e tet et (k 1)! (i) ( A) C x(t ) 0 as t (ii) ( A) with C x(0) x(t ) (iii) ( A) C j axisand each eigenvalue at jw-axis has index 1 All solution are bounded. (iv) (A) C- j axis and jw-axis having index > 1 solution which is not bounded. 長庚大學電機系 4-11 Discretization Consider x Ax Bu y Cx Du Suppose that u(t ) u(kT ) for kT t (k 1)T Define u[k ]: u(kT ) kT AkT x[k ] : x( kT ) e x(0) e A( kT ) Bu ( ) d 0 x[k 1] e A ( k 1)T x(0) ( k 1)T 0 e x[k ] AT ( k 1)T kT e A( kT T ) Bu ( ) d e A( kT T ) Bu ( )d T e x[k ] ( e A d ) Bu[k ] AT 0 (by letting kT T and the assumption on u(t )) 長庚大學電機系 4-12 Thus, if input changes value only at discrete-time instance kT and we compute only the responses at t=kT then system x Ax Bu y Cx Du becomes x[k 1] Ad x[k ] Bd u[k ] y[k ] Cd x[k ] Dd u[k ] with T Ad e , Bd ( e A d ) B, Cd C, Dd D AT T Bd ( e A d ) B 0 0 T ( ( I A ...)d ) B 0 (TI AT 2 / 2! ...) B A1 (e AT I ) B if A is nonsingular 長庚大學電機系 4-13 Solution of Discrete-time Equation x[k 1] Ax[k ] Bu[k ] y[k ] Cx[k ] Du[k ] Consider Note that, x[1] Ax[0] Bu[0] x[2] Ax[1] Bu[1] A2 x[0] ABu[0] Bu[1] In general, k 1 x[k ] A x[0] Ak 1m Bu[m] k m0 k 1 y[k ] CAk x[0] CAk 1m Bu[m] Du[k ] m 0 長庚大學電機系 4-14 Stability of Zero-Input Response Let A QJQ1 Am QJ mQ1 m J km ( ) 0 m (m k 2) m k 1 (k 1)! , m k 1 m If i ( A) 1, i Am x[0] 0, as m , x[0]. If i i ( A) 1 or i ( A) 1 with index 1 Am x[0] is unbounded for some x[0] as m goes to infinity. 長庚大學電機系 4-15 Equivalent State Equations To transform the system into a canonical form or as simple as possible. Consider x Ax Bu y Cx Du Let x Px, P is a nonsingular matrix x Ax Bu y Cx Du A PAP1, B PB, C CP1, D D the two systems are called equivalent and x Px is called an equivalence transformation. 長庚大學電機系 4-16 ( ) det( I A) det( PP 1 PAP 1 ) det( P) det( I A) det( P 1 ) det( I A) ( ) i.e., equivalent state equation have the same characteristic polynomial and thus the same set of eigenvalues. Gˆ C ( sI A) 1 B D CP 1[ P ( sI A) P 1 ]1 PB D CP 1 P ( sI A) 1 P 1 PB D C ( sI A) 1 B D Gˆ ( s ) i.e., transfer matrices are preserved under equivalence transformation. 長庚大學電機系 4-17 Two state equation are said to be zero-state equivalent if they have the same transfer matrix. In such a case, D C (sI A)1 B D C (sI A)1 B D CA Bs D CAi 1Bs i i 1 i i 1 i 1 Theorem: Two LTI system {A,B,C,D} and {A, B, C, D} are zero-state equivalent (or have the same transfer matrix if and only if D D and CAi B CAi B, i 0,1,... Clearly, two equivalent LTI systems implies that they are zero-state equivalent. 長庚大學電機系 4-18 Realization 1 Question: Given (A,B,C,D) Gˆ (s) C(sI A) B D . Conversely, given Gˆ (s) and suppose that the LTI system is lumped, how to find state-space representation (A,B,C,D)? ˆ A transfer matrix G( s) is said to be realizable if there exist {A,B,C,D} such that Gˆ (s) C(sI A)1 B D {A,B,C,D} is called a realization of Gˆ (s). If the realization problem is solvable, then the state space description is not unique (discussed later) 長庚大學電機系 4-19 Theorem: A q p transfer matrix Gˆ (s) is realizable , each entry of Gˆ (s) is a proper rational function. 長庚大學電機系 4-20 The realization discussed above (in the proof of the last theorem) is called controllable canonical form. A dual realization called observable canonical form is described below: 1 I q 2 I q A r 1 I q I r q C Iq 0q Iq 0q 0q Iq 0q 0q 0q 0q 0q N1 0q N2 , G Iq Nr 0q 0q , D Gˆ () 長庚大學電機系 4-21 Solutions of LTV Equations Consider x(t ) A(t ) x(t ), A(t ) Rnn 0 A(t ) C Solution ! for any given initial state. 0 We assume hereafter that A(t ) C The solution set is a vector space of dimension n X (t ) : ( x1 (t ),..., xn (t )) Rnn is called a fundamental matrix of x(t ) A(t ) x(t ) if each column xi (t ) is a solution, and {x1 (t )...xn (t )} is linearly independent. Note that, there are infinite many fundamental matrices 長庚大學電機系 4-22 The fundamental matrix (t, t0 ) X (t ) X 1 (t0 ) is called the state transition matrix. (t , t0 ) is also a fundamental matrix. (t , t0 ) is unique no matter what fundamental matrix X (t ) is chosen. (Reason: Let Y (t ) X (t )C Y (t )Y 1 (t0 ) X (t ) X 1 (t0 )) (t , t ) I (t, t0 )(t0 , ) (t, ) The unique solution of 1 (t, t0 ) (t0 , t ) x A(t ) x is (t , t0 ) x0 x(t0 ) x0 長庚大學電機系 0 0 x(t ) or Example: consider x(t ) t 0 x1 (t ) 0 x1 x1 (0) x1 (t ) 0 x2 (t ) tx1 (t ) 1 x2 (t ) tx1 x2 (t ) x1 (0)t 2 x2 (0) 2 1 1 x(t ) 1 2 is a solution with x(0) t 0 2 0 0 x(t ) is a solution with x(0) 1 1 0 1 is a fundamental X (t ) 2 t / 2 1 4-23 matrix since it has linearly independent columns. 1 (t , t0 ) X (t ) X 1 (t0 ) 1 2 t 2 1 0 1 1 0 1 2 t 10 1 2 0 1 長庚大學電機系 4-24 Let (t, t0 ) be the state transition matrix of x(t ) A(t ) x(t ). Then the unique solution of x(t ) A(t ) x(t ) B(t )u(t ) is y(t ) C (t ) x(t ) D(t )u (t ) t x(t ) (t , t0 ) x(t0 ) (t , ) B( )u ( )d t0 y (t ) C (t )(t , t ) x(t ) C (t ) t (t , ) B( )u ( )d D(t )u (t ) 0 0 t0 zero-input response zero state response t yzs (t ) [C (t )(t , ) B( ) D(t ) (t )]u ( )d t0 G(t , ) C (t )(t , ) B( ) D(t ) (t ) At A If A ñ constant matrix (t, ) e e e A(t ) the unique solution of x(t ) Ax(t ) Bu(t ) is t A(t t0 ) x(t ) e x(t0 ) e A(t ) Bu( )d t0 長庚大學電機系 4-25 Discrete-Time Case x[k 1] A[k ]x[k ] B[k ]u[k ] Consider y[k ] C[k ]x[k ] D[k ]u[k ] x[k0 2] A[k0 1]x[k0 1] B[k0 1]u[k0 1] A[k0 1] A[k0 ]x[k0 ] A[k0 1]B[k0 ]u[k0 ] B[k0 1]u[k0 1] x[k0 3] A[k0 2] A[k0 1] A[k0 ]x[k0 ] (k0 3, k0 ) (k0 3, k0 1) A[k0 2]A[k0 1]B[k0 ]u[k0 ] A[k0 2]B[k0 1]u[k0 1] B[k0 2]u[k0 2] (k0 3, k0 2) 長庚大學電機系 4-26 For k k0 define [k 1, k0 ] A[k ][k , k0 ], [k0 , k0 ] I . [k , k0 ] A[k 1] A[k0 ], for k k0 [k , k0 ] [k , k1 ][k1, k0 ], for k k1 k0 The solution of x[k 1] A[k ]x[k ] B[k ]u[k ] is x[k ] [k , k0 ]x[k0 ] k 1 [k , m 1]B[m]u[m] m k0 for k k0 If we define [k , k0 ] 0 for k k0 y[k ] C[k ]x[k ] D[k ]u[k ] C[k ][k , k0 ]x[k0 ] k (C[k ][k , m 1]B[m] D[m] [k m])u[m] m k0 G[ k ,m ] 長庚大學電機系 4-27 Equivalent Time-Varying Equation x A(t ) x B(t )u Consider .........................................(V1) y C (t ) x D(t )u Let x P(t ) x. Assume that P(t )is nonsingular, P(t ) and are continuous. x A(t ) x B (t )u .........................................(V2) y C (t ) x D(t )u 1 where A(t ) [ P(t ) A(t ) P(t )]P (t ) B (t ) P(t ) B(t ) C (t ) C (t ) P 1 (t ) D(t ) D(t ) The two systems are called equivalent. And P(t ) is called an equivalence transformation. 長庚大學電機系 4-28 Recall that System (V1) has impulse response G(t, ) C(t ) X (t ) X 1 ( ) B( ) D(t ) (t ) The impulse response for System (V2) is G (t , ) C (t ) X (t ) X 1 ( ) B ( ) D(t ) (t ) C (t ) P 1 (t ) P(t ) X (t ) X 1 ( ) P 1 ( ) P( ) B( ) D(t ) (t ) G (t , ) Thus, the impulse response matrix is invariant under any equivalence transformation. 長庚大學電機系