Controllability and Observability Controllability and observability are important structural properties of dy- namic system. First identified and studied by Kalman (1960) and later by Kalman ct al. (1962), these properties have continued to be examined dur- ing the last four decades, We will discuss only a few of the known results for linear constants systems with one input and one output, In the text we dis- ccuss these concepts in connection with control law and estimator designs. For ‘example, in Section 7.2 we suggest that if the square matrix given by GGG... PG] (Dy) is nonsingular, then by transformation of the state we can convert the given description into control canonical form. We can then construct a control law that will give the closed-loop system an arbitrary characteristic equation. 849)850 Appendix D Controllability and Observability Detnton 1 Detintion D.1_ Controllability ‘We begin our formal discussion of controllability with the first of four defi tions: The system (F, G) is controllable if for any given nth-order polynomial ae(s) there exists a (unique) control aw u = Kx such that the characteristic poly- nomial of F ~ GK is @(s) From the results of Ackermann’s formula (see Appendix E), we have the fol- lowing mathematical test for controllability: (F, G) is a controllable pair if and only if the rank of © isn. Definition I based on pole placement is a frequency-domain concept. Controllability can be equivalently defined in the time domain, ‘Thesystem (F,G)iscontrollableif there existsa (piecewise continuous) control signal u(2) that will take the state of the system from any initial state Xy to any desired final state Xp ina finite time interval We will now show that the system is controllable by this definition if and only if @ is full-rank, We first assume that the system is controllable but that rank[G FG F°G ... F'G) 0. This implies that all points reachable from the origin are ‘orthogonal to ¥. This restricts the reachable space and therefore contradicts, the second definition of controllability. Thus if @ is singular, (F, G) is not controllable by Definition IL Next we assume that @ is full-rank but (F, G) isuncontrollable by Definition IL. This means that there exists a nonzero vector v such that a FY -OGu(r jdt = 0, (p10) because the whole state space is not reachable. But Eq. (D.10) implies that veFy-9G, . Oe sty (pa If we sot r = ty, we see that vG = 0. Also, differentiating Eq. (D.11) and letting r = ry gives vEG = 0, Continuing this process, we find that VG = vFG = PG = (p12) which contradicts the assumption that © is We have now shown that the system is controllable by Definition IT if and only if the rank of € is , exactly the same condition we found for pole assignment. Our third definition comes closest to the structural character of controllability: ‘The system (F, G) is controllable if every mode of Fis connected to the control input. Because of the generality of the modal structure of systems, we will only treat the case of systems for which F can be transformed to diagonal form. (The Gouble-integration plant does nor qualify.) Suppose we have a diagonal matrix F, and its corresponding input matrix G, with elements g;. The structure of such a system is shown in Fig, D1. By definition, for a controllable system the input must be connected to each mode so that the gare all nonzero. However, this is not enough if the poles (3) are not distinct. Suppose, for instance, that 1, = 2a. The first two state equations are then ug = Aisa + gu fing = Auaa + go (D113)852 Appendix D Figure D.1 Blok agra fa system -] 5 wi 2 dagona mae | Controllability and Observability If we define a new state, é f2k1d — grag, the equation for & is goiig — giiag = grhuy + gre — give — gi@M= AE, — (D14) which does not include the control uw; hence & is not controllable. The point is that if any two poles are equal in a diagonal Ey system with only one input, we effectively have a hidden mode that is not connected to the control, and the system is not controllable (Fig, D.2(a)]. This is because the two state variables move together exactly, so we cannot independently control x14 and x2. There- fore, even in such a simple case, we have two conditions for controllability 1. Alleigenvalues of Fy are distinct 2. No element of Gy is zero, Now let us consider the controllability matrix of this diagonal system. By direct computation, a1 gh eu] eh Br Bahn : oft a at = Lad (p15) Figure D2 Examples of uncontolabe systems Section D1 Controllability 853 Note that the controllability matrix @ is the product of two matrices and is nonsingular if and only if both of these matrices are invertible. The first matrix has a determinant that is the product of the g;, and the second matrix (called 1a Vandermonde matrix) is nonsingular if and only if the 4, are distinct. Thus Definition 111 is equivalent to having a nonsingular € also. Important o the subject of controllability is the Popoy-Hautus-Rosenbrock (PHR) test (see Rosenbrock, 1970, and Kailath, 1980), which is an alternate way to test the rank (or determinant) of @. The system (F, G) is controllable if the system of equations v[sI-F G]=0" (D116) has only the trivial solution v? = 0", that is, if the following matrix pencil is full-rank for all s rank[sI—F G] =n, (D7) or if there is no nonzero v" such that! WE =sy!, (D8) VG=0. (D.19) ‘This test is equivalent to the rank-of-C test. It is easy to show that if such a vector v exists, then @ is singular. For, if a nonzero ¥ exists such that ¥"G = 0, then by Egs. (D.18) and (D.19) we have WRG = sv"G =0. (D20) Then, multiply 1g by FG, we find that WEG = sv FG = 0, (p21) and so on, ‘Thus we determine that v7 = 0” has a nontrivial solution, that C is singular, and that the system is not controllable. To show that a nontrivial v? exists if € is singular requires more development, which we will not give here (see Kailath, 1980). We have given two pictures of uncontrollability. Either a mode is phys- ically disconnected from the input [Fig. D.2(b)], or else two parallel subsys- tems have identical characteristic roots (Fig, D.2(a)]. The control engineer should be aware of the existence of a third simple situation, illustrated in Fig. D.2(¢) namely, a pole-zero cancellation. Here the problem is that the ‘mode at s = 1 appears to be connected to the input but is masked by the zero at s = 1 in the preceding subsystem; the result is an uncontrollable system. "Visa et eigenvector of F.854 Appendix D Controllability and Observability Detinten This can be confirmed in several wa matrix. The system matrices are eth 6 [3] so the controllability matrix is 198. First let us look at the controllability [c rej=[) ae (ox which s clearly singular. The controllability matrix may be computed using the cro command in Matta fe) = cto). If we compute the transfer function from w to 9, we find 1 1 : . 23 wo Fa(=a) =a mas Because the natural mode at s = I disappears from the input-output deserip- tion, itis not connected to the input. Finally, if we consider the PHR test, [wr al=['! a 7] (na and let s = 1, then we must test the rank of 2 0 -2 -101 which is clearly less than 2. This result means, again, that the system is uncontrollable. ‘The asymptotically stable system (F, G) is controllable if the controllability Gramian, the square symmetric matrix €,. given by the solution o the ollow- ing Lyapunov equation Fe, +6,F' +66" 6. (p25) isnonsingular. The controllability Gramianis also the solution tothe following integral equation c= [Fete a ox geek nent ee Section D2 Observability 855 In conclusion, the four definitions for controllability—pole assignment (Defini- tion 1), state reachability (Definition 11), mode coupling to the input (Definition 111),and controllability Gramian (Definition IV)—are equivalent. The tests for any of these four properties are found in terms of the rank of the controllability or controllability Gramian matrices or the rank of the matrix pencil [sl - F G]. 1F€ is nonsingular, then we ean assign the closed-loop poles arbitrarily by state feedback, we can move the state to any point in the state space in a finite time, ‘and every mode is connected to the control input.’ We have shown the latter for diagonal F only, but the result is true in general D.2 Observabilty So far we have discussed only controllability. ‘The concept of observability is parallel to that of controllability, nd all ofthe results we have discussed thus far ‘may be transformed to statements about observability by invoking the property Cf duality, a discussed in Section 7.5.1, The observability definitions analogous {0 those for controllability are as follows: 1. Definition I. The system (FH) is observable if, for any nth-order polynomial (3), there exists an estimator gain L such that the characteristic equation of the state estimator error is a, (8). 2, Definition 11. The system (F, H) is observable if, for any x(0), there isa finite time + such that x(0) can be determined (uniquely) from u(t) and y(9) for Osrer Definition II. The system (F, H) is observable if every dynamic mode in F is connected to the output through H. 4, Definition IV. The asymptotically stable system (F, H) is observable if the observability Gramian is nonsingular. As we saw in the discussion for controllability, mathematical tests can be de- veloped for observability. The system is observable if the observability matrix H HF o= (p27) Her! is nonsingular, If we take the transpose of 6 and let H” = G and ET =F. then ‘we find the controllability matrix of (F, G), another manifestation of duality The observability matrix, ©, may be computed using the obsv command in ‘We have shown the latter for diagonal F only, but the cesut ste in general6 Appendix DD Controllabili and Observability Marta [oo] = obsv(F H). The system (F, H) is observable ifthe following matrix pencil is full-rank for all s: wet =m wx) ‘The observability Gramian, O,, which is a symmetric matrix, and the solution to the integral equation, [ FTO, +0,F +H" =0, (D30) command in Marcas (og) = gram(F",H’). The observability Gramian has an in- terpretation as the “information matrix” in the context of estimation he Ackermann’s Formula for Pole Placement Given the plant and state-variable equation = Fx + Gu, (1) ‘our objective is to find a state feedback control law. w= Kx (E2) such that the closed-loop characteristic polynomial is, ac(s) = det(st ~ F + GK), (3) First we have toselect ae(s), which determines where the polesare to be shifted, then we have to find K such that Eq. (E.3) will be satisfied. Our technique is based on transforming the plant equation into control canonical form. We begin by considering the effect of an arbitrary nonsingular transforma tion of the state, x=, (Ea) where & is the new transformed state. The equations of motion in the new coordinates, from Eq. (E.4), are X= TR = Fx + Gu = FIX + Gu. (ES) +6u (E86) 897