Anda di halaman 1dari 7

146 Asian Journal of Control, Vol. 1, No. 3, pp.

146-152, September 1999

BLOCK DECOUPLING CONTROL OF LINEAR


MULTIVARIABLE SYSTEMS
Ching-An Lin and Chang-Ming Wu

ABSTRACT

This paper gives parameterizations of block decoupling controllers and


achievable block-diagonal I/O maps for linear multivariable plants under
unity-feedback configuration. The parameterizations, based on coprime
factorizations and a special generalized Bezout identity, do not require the
computation of Smith McMillan forms and, thus, can be reliably computed. It
is also shown that through the construction of an admissible block decoupling
precompensator, block decoupling controller design reduces to a set of con-
troller designs for the block channels which are smaller in dimension.

Key Words: Block decoupling, linear multivariable systems, precompensator.

I. INTRODUCTION decoupling controller can be designed by first block


decoupling the plant by open-loop precompensation and
Problems related to block decoupling control of then designing feedback controllers for each block channel.
linear multivariable systems have been studied by many A block decoupling precompensator may not maintain
authors. Commault et al. [3] considered block decoupl- stabilizability due to CI+ pole-zero cancellations [1]; even
ing by a combination of dynamic precompensation and if stabilizability is maintained, the precompensator may
state feedback and gave a lower bound on the McMillan impose additional constraints on the achievable block
degree of the achievable I/O transfer matrices. Under diagonal I/O maps. In this paper, we construct a block
unity-feedback configuration, Linnemann and Wang decoupling precompensator which maintains stabiliza-
[7] gave conditions for the solvability of the block bility and puts no additional constraint on the achievable
decoupling problem and a parameterization of all block block diagonal I/O maps. With this precompensator, the
decoupling controllers based on the notion of skew- block decoupling controller design essentially reduces to a
coprimeness and stability factorization. Lin and Wu [6] set of controller designs for block channels which are
gave necessary and sufficient conditions for the existence smaller in dimension.
of block decoupling controllers. Desoer and Gundes [4] This paper is organized as follows. Basic definitions
gave the design of a decoupling compensator for linear and preliminary results are given in Section 2. Section 3
multivariable systems. gives parameterizations of all block decoupling controll-
In this paper, we will give parameterizations of all ers and of all achievable block diagonal I/O maps. Section
block decoupling (stabilizing) controllers and all 4 describes the construction of a block decoupling precom-
achievable block diagonal I/O maps. The parameteriza- pensator and shows that it has the desired properties.
tions are based on coprime factorizations of the plant and Section 5 is a brief conclusion.
a special generalized Bezout identity. In addition to the
computation of coprime factorization, only computation of Notation: IR := the field of real numbers; CI := the field
the inverses of proper rational matrices is required [5]. of complex numbers; CI_ := {s ∈ CIRe(s) < 0}; CI+ :=
These computations can be more reliably done than {s CIRe(s) ≥ 0}; IR(s) := the set of rational functions in s
the computation of the Smith McMillan form of the plant, with real coefficients; IRp(s) ( IRpo(s), respectively) :=
which was required in [3] and [7]. Conceptually, a block the set of proper (strictly proper, respectively) rational
functions in s with real coefficients. For H(s) ∈ IR(s)n × m,
Manuscript received April 2, 1999; revised and accepted June
Z[H] := the set of all zeros of H in CI, P[H] := the set of
22, 1999. all poles of H in CI [2], Z+[H] := Z[H] ∩ CI+ and P+[H] :=
The authors are with Dept. of Electrical and Control P[H] ∩ CI+. S := {h ∈ IRp(s)h has no pole in CI+}. A list
Engineering, National Chiao-Tung University, Hsinchu, Taiwan. of positive integers (n1, n2, …, nk) satisfying Σ ki = 1 ni = n
This work was supported by the National Science Council of is said to be a partition of n. We use diag{H i} ki = 1 to
the Republic of China under Grant NSC-87-2213-E-009-144. denote the block diagonal rational matrix with Hi as its
C.-A. Lin and C.-M. Wu: Block Decoupling Control of Linear Multivariable Systems 147

ith block diagonal entry, where H i ∈ IR(s)ni × ni. For are left coprime and D –pl1 Npl = P. We need the following
f, g ∈ IR[s], deg(f) := degree of f, the relative degree of assumptions:
f/g ∈ IR(s) is defined as deg(f) – deg(g). For H = [hij] ∈
IR(s)n × m, the relative degree of H is defined as the largest A1) The plant P has full normal row rank, and
relative degree of hij, 1 ≤ i ≤ n and 1 ≤ j ≤ m. Two matrices A2) P+[P] ∩ Z+[P] = Ø.
A ∈ S n × n and B ∈ S n × n are equivalent if there exist
S-unimodular matrices U and V such that A = UBV. Assumption A1) implies that m ≥ n, and it is neces-
sary for block decoupling [7]; assumption A2) means
II. PRELIMINARY that the CI+-poles and CI+-zeros of P do not coincide.
These assumptions together are sufficient for the exist-
Consider the unity-feedback system S(P, C) shown in ence of block decoupling controllers for P. Although A2)
Fig. 1, where P ∈ IRpo(s)n × m is the plant, C ∈ IRp(s)m × n is is not necessary [6], it is generically satisfied, and it well
the controller, (u1, u2) is the input and (y1, y2) is the output. simplifies the derivations to be described below.
Let u := [u 1T u 2T ] T and y := [y1T y2T ] T . The closed-loop
transfer matrix Hyu ∈ IRp(s)(n + m) × (n + m) is given by III. PARAMETERIZATIONS OF BLOCK
DECOUPLING CONTROLLERS
H y1u1 H y1u2 C(I + PC) – 1 – CP(I + CP) – 1 Consider the plant P ∈ IRpo(s)n × m, which is a strictly
H yu = = .
H y2u1 H y2u2 PC(I + PC) – 1 P(I + CP) – 1 proper rectangular rational matrix. Since P ∈ IRpo(s)n × m
has full normal row rank, by assumption A1), there exists
(2.1)
a unimodular matrix U ∈ Sm × m such that
Definition 2.1. The system S(P, C) is stable if and only if
PU = [P 0], (3.1)
Hyu is stable. We say that C stabilizes P or C is a stabilizing
controller for P if S(P, C) is stable.
where P ∈ IRpo(s)n × m is nonsingular. Clearly P satisfies
assumptions A1) and A2). Let (Nr, Dr) be a right coprime
Definition 2.2. The system S(P, C) is block decoupled
factorization of P, and let (Dl, Nl) be a left coprime factor-
with respect to the partition (n1, …, nk)1 if C stabilizes P and
ization of P, where Nr, Dr, Nl, Dl ∈ Sn × n. We note that
the I/O map Hy2u1 is nonsingular and block diagonal with
these coprime factorizations can be computed through
respect to the partition (n1, …, nk). We say that C is a
standard state space procedure [8]. Let
block decoupling controller for P if S(P, C) is block
decoupled with respect to the partition (n1, …, nk).
Dr 0
N pr = N r 0 , D pr = U , (3.2)
Remark. We note that decoupling is a special case of 0 I
block decoupling with each block size ni = 1.
N pl = N l 0 U – 1, D pl = D l , (3.3)
n×n
Definition 2.3. A stable rational matrix M ∈ S is an
achievable I/O map of the system S(P, C) if there exists a where Npr, Npl ∈ Sn × m, Dpr ∈ Sm × m and Dpl ∈ Sn × n. Then,
stabilizing controller C such that Hy2u1 = M. (Npr, Dpr) and (Dpl, Npl) are, respectively, a right coprime
factorization of P and a left coprime factorization of P. In
We say that (Npr, Dpr) is a right coprime factorization the following, we construct a special generalized Bezout
of P if Npr ∈ Sn × m and Dpr ∈ Sm × m are right coprime and identity which is suitable for the description of block
NprD –pr1 = P. Similarly, we say that (D pl, Npl) is a left decoupling controllers. We can write
coprime factorization of P if Dpl ∈ Sn × n and Npl ∈ Sm × m

Z ij
N r– 1 = , (3.4)
Pij–Pij+

where Zij, Pij–, Pij+ ∈ IR[s] are mutually coprime, Pij+ is


monic, Z[Pij+] ⊂ CI+ and Z[Pij–] ⊂ CI–. We choose λ ∈ S
such that the numerator polynomial of λ is the monic
least common multiple of {Pij+} i,n j = 1, and the relative degree
of λ–1 is the same as that of N r– 1 . Similarly, we write
Fig. 1. Unity-feedback system S(P, C).

N ij
1
We consider block decoupling with respect to a fixed but D l– 1 = , (3.5)
arbitrary partition (n1, …, nk) throughout.
D ij–D ij+
148 Asian Journal of Control, Vol. 1, No. 3, September 1999

where Nij, Dij–, Dij+ ∈ IR[s] are mutually coprime, Dij+ is


monic, Z[D ij+] ⊂ CI+ and Z[D ij–] ⊂ CI–. We can then Q1
choose ψ ∈ S such that the numerator polynomial of ψ is HP = N pr U cl + D Q 1 ∈ S n × n and Q 2 ∈ S (m – n) × n
Q 2 pl
the monic least common multiple of {D ij+} ni, j = 1 , and the
relative degree of ψ–1 is the same as that of D l– 1 . Since (3.11)
Dl(∞) is nonsingular, D l– 1 is proper but not strictly proper.
Thus, ψ(∞) ≠ 0. It then follows that λN r– 1 ∈ Sn × n and ψD l– 1 = {αλI + NrQ1Dl|Q1 ∈ Sn × n} (3.12)
∈ Sn × n. The zeros of λ and ψ define the CI+ zeros of Nr
and Dl, respectively [2]. Since Nr and Nl are equivalent, the by (3.2), (3.3) and (3.8).
zeros of λ are also the CI+ zeros of Nl, and the relative The set of achievable block diagonal I/O maps is a
degree of N l– 1 is the same as that of N r– 1 . Thus, λN l– 1 ∈ subset of HP defined in (3.12). Since αλI is diagonal,
Sn × n. Similarly, ψD r– 1 ∈ Sn × n. Moreover, p0 ∈ P+[P] if stable and proper, it follows from (3.12) that the set of all
and only if ψ(p0) = 0; z0 ∈ Z+[P] if and only if λ(z0) = achievable block diagonal I/O maps Hy2u1 is
0. By assumption A2) and that ψ(∞) ≠ 0, λ and ψ have no
common zeros in CI+ ∪ {∞}; hence, λ and ψ are coprime. HPd = {αλI + NrQ1Dl|Q1 ∈ Sn × n
Thus, there exist α, β ∈ S such that
and NrQ1Dl is block diagonal}. (3.13)
αλ + βψ = 1. (3.6)
Also, from (3.7), (3.8) and (3.10), the set of all block
Define decoupling controllers for P is

αλN l– 1
U cr = , V cr = βψD l– 1, (3.7)
0
αλN l– 1 + D rQ 1 –1
C Pd = U βψD l– 1 – N rQ 1
Q2
αλN r– 1 βψD r– 1 0
U cl = , V cl = U – 1 , (3.8)
0 0 I

where Ucr, Ucl ∈ Sm × n, Vcr ∈ Sn × n, and Vcl ∈ Sm × m. It is


easy to check that they satisfy the following generalized 1 Q 1 ∈ S n × n, Q 2 ∈ S (m – n) × n and
. (3.14)
Bezout identity: N rQ 1D l is block diagonal

V cl U cl D pr – U cr
= I0 . (3.9)
– N pl D pl N pr V cr 0I
In the descriptions of HPd and CPd, (3.13) and (3.14),
respectively, the parameter Q1 is not free but must satisfy
The set of all stabilizing controllers for P is, thus, that NrQ1D l is block diagonal. In order to obtain an
given by, unconstrained parameterization, we partition N r– 1 as

N r– 1 = N 1 Nk , (3.15)
–1
Q1 Q1
CP = U cr + D pr V cr – N pr
Q2 Q2 where Ni ∈ IR(s)n × ni, 1 ≤ i ≤ k. Let (Γi, Θi) be a right
coprime factorization of Ni, 1 ≤ i ≤ k, where Γi ∈ Sn × ni
and Θi ∈ Sni × ni. Thus, N r– 1 has the following property:

N r– 1 = Γ 1Θ –1 1 Γ kΘ –k 1 = ΓΘ – 1 , (3.16)
1

Q 1 ∈ S n × n and Q 2 ∈ S (m – n) × n . (3.10)
where Γ = [Γ1 … Γk] and Θ = diag{Θ i} ki = 1. Since, for each
i, P+[Ni] ⊂ P+[N r– 1 ] = Z+[P] and P+[Ni] = Z+[Qi], it follows
that
Since NprUcl = αλI, the set of all achievable I/O maps
Hy2u1 is Z+[Θ] ⊂ Z+[P]. (3.17)
C.-A. Lin and C.-M. Wu: Block Decoupling Control of Linear Multivariable Systems 149

Similarly, we can partition D l– 1 as Q 1 = N r– 1ΘQ d∆D l– 1 = ΓQ dΩ , (3.22)


T
D l– 1 = D 1T D kT , (3.18) where we have used (3.16) and (3.19), it follows from
(3.14) that the set of all block decoupling controllers is

where Di ∈ IRp(s)ni × n, 1 ≤ i ≤ k. Let (∆i, Ωi) be a left


coprime factorization of Di, 1 ≤ i ≤ k, where Ωi ∈ Sni × n,
∆i ∈ Sni × ni and det ∆i(∞) ≠ 0. Let ∆ = diag{∆ i} ki = 1 and αλN l– 1 + D rΓQ dΩ –1
Ω = [Ω1T … ΩkT ]T. Then, C Pd = U βψD l– 1 – N rΓQ dΩ
Q2
T
D l– 1 = (∆ –1 1Ω1) T (∆ –k 1Ωk) T = ∆ – 1Ω .
(3.19)

For 1 ≤ i ≤ k, P+[Di] ⊂ P+[D l– 1 ] = P+[P] and P+[Di] 1 Q d is block diagonal and stable
= Z+[∆i]. It then follows that . (3.23)
Q 2 ∈ S (m – n) × n
Z+[∆] ⊂ P+[P]. (3.20)

The following lemma is used in the proof of


Theorem 3.2, which yields a free parameter description of Proof. Let Q1 ∈ Sn × n be such that H := NrQ1Dl is block
HPd and CPd. diagonal. Since N r– 1 H = Q1Dl is stable and proper, by
Lemma 3.1, it follows from (3.16) that Θ–1H is stable
Lemma 3.1. Suppose H = diag{H i} ki = 1 ∈ Sn × n is block and proper. Similarly, since HD l– 1 = NrQ1 is stable and
diagonal, and that Θ and ∆ are as defined in (3.16) proper, it follows from (3.19) that H∆–1 is stable and
and (3.19), respectively. Then, (i) N r– 1 H ∈ Sn × n if and only proper. Let Qd = Θ–1H∆–1, which is block diagonal; we
if Θ–1H ∈ Sn × n, and (ii) HD l– 1 ∈ Sn × n if and only if H∆–1 ∈ then have H = ΘQ d∆. It then suffices to show that
Sn × n. Θ–1H∆–1 is stable and proper. Since H∆–1 and Θ–1H are
stable, P+[Θ–1H∆–1] ⊂ P+[Θ–1] and P+[Θ–1H∆–1] ⊂ P+[∆–1].
Proof. Since H is block diagonal, N r– 1 H ∈ Sn × n if and Consequently, P +[Θ –1 H∆ –1] ⊂ (P +[Θ –1] ∩ P +[∆ –1 ]) =
only if ΓiΘ –i 1 Hi ∈ Sn × ni, 1 ≤ i ≤ k. Since Θi and Γi are (Z+[Θ] ∩ Z+[∆]). By (3.17), (3.20) and assumption A2),
right coprime, ΓiΘ –i 1 Hi ∈ Sn × ni if and only if Θ –i 1 Hi ∈ P+[Θ–1H∆–1] ⊂ (Z+[P] ∩ P+[P]) = Ø; thus, Θ–1H∆–1 is stable.
Sni × ni [9] for 1 ≤ i ≤ k. Thus, N r– 1 H ∈ Sn × n if and only if Since Θ–1H and ∆–1 are proper, Θ–1HD–1 is proper, and the
Θ–1H ∈ Sn × n. This proves (i). The proof of (ii) is similar. proof is complete. ■

The equations (3.21) and (3.23) completely param-
n×n
Theorem 3.2. Suppose the plant P ∈ IRpo(s) has full eterize the set of achievable block decoupled I/O maps and
normal row rank and has no CI+-coincidence poles or the set of block decoupling controllers via a block diagonal
zeros. Let (Nr, Dr) and (Dl, Nl) be, respectively, the right parameter. Thus, the design of the block decoupling
and left coprime factorizations of P. If Qi ∈ Sn × n is such controller becomes the task of choosing two appropriate
that NrQ1Dl is block diagonal, then there is a Qd ∈ Sn × n parameters so that the design requirements are satisfied. If
block diagonal such that ΘQd∆ = NrQ1Dl, where the block the requirements are specified on the I/O map and the
diagonal matrices Θ and ∆ are defined, respectively, through sensitivity map (where they are both block diagonal), then
(3.16) and (3.19). the design reduces to a set of k block channel designs of
smaller dimension. We emphasize again that the param-
Comments. (a) Since Γ and Ω are stable and proper, eterization given above does not require computation of the
ΓQ dΩ is stable and proper for every Qd ∈ S n × n block Smith-McMillan form of the plant; hence, it can be reliably
diagonal. Let Q1 = ΓQdΩ ∈ Sn × n, then, NrQ1Dl = ΘQd∆ computed.
is block diagonal. It then follows from the theorem and
(3.13) that the set of all achievable block decoupled I/O IV. BLOCK DECOUPLING
maps Hy2u1 for the plant P is PRECOMPENSATORS

HPd = {αλI + ΘQd∆|Qd is block diagonal and stable}. Another way to design block decoupling controllers
(3.21) is to first find a block decoupling precompensator which
makes the transfer matrix of the cascade connection (of the
(b) Since NrQ1Dl = ΘQd∆ if and only if precompensator and the plant) block diagonal and then
150 Asian Journal of Control, Vol. 1, No. 3, September 1999

design feedback controllers for each of the block channels.


Two problems, however, may arise in this approach. The
cascade connection may not be stabilizable and the block
decoupling precompensator may introduce additional
constraints on the achievable block decoupled (closed-
loop) I/O maps.
Fig. 2. The feedback system S(P, F , A).
We say the block decoupling precompensator F
for the plant P is admissible if the cascade connection of
F and P is stabilizable. In this section, we will construct
an admissible block decoupling precompensator for
P which does not introduce any constraint on the achiev-
able block decoupled I/O maps. First, we construct
an admissible block decoupling precompensator F for
P ∈ IRpo(s)n × n given in (3.1) that is nonsingular and
satisfies assumption A2). In order to find a block decoupl-
ing precompensator for P, we need the following. Fig. 3. Unity-feedback system S(PF, A).

Proposition 4.1. Let λ, ψ, Θi and ∆i be as defined in the


previous section. The following holds: FA(I + PFA) – 1 – FAP(I + FAP) – 1 F(I + APF) – 1
H yu = PFA(I + PFA) – 1 P(I + FAP) – 1 PF(I + APF) – 1 .
(i) λΘ –i 1 ∈ S n i × n i and ψ∆ –i 1 ∈ S n i × n i , and A(I + PFA) –1
– AP(I + FAP) – APF(I + APF) – 1
–1

(ii) λΘ –i 1 and ∆i are right coprime.


(4.3)
Proof. (i) Since λN = [λΓ Θ … λΓkΘ ] ∈ S and
–1
r 1 1
–1 –1
k
n×n

Now, suppose the block diagonal controller A is


Γi and Θi are right coprime, λΘ ∈ Sni × ni. Similarly, ψ∆ –i 1
–1
i
such that the system S(PF, A) is stable. It is easy to verify
∈ Sni × ni. (ii) By (3.6), we have αλI + βψI = I, where α, β
that the following Bezout identity
∈ S. The matrices αΘi, λΘ –i 1 , βψ∆ –i 1 and ∆i are stable and
proper. Since (αΘi) (λΘ –i 1 ) + (βψ∆ –i 1 )∆i = αλI + βψI = I,
βψ∆ αλΘ – 1 ∆ – αΘ
–1
λΘ –i 1 and ∆i are right coprime, 1 ≤ i ≤ k. ■ = I0 (4.4)
– λΘ
–1
∆ Θ βψ∆ – 1 0I
Since det ∆i(∞) ≠ 0, (λΘ –i 1 )∆ –i 1 ∈ IRp(s)ni × ni. There
exists Θ i , ∆ i ∈ Sni × ni left coprime such that
follows from (3.6) and (4.1), and that each element in
(4.4) is stable and proper. Since Θ and ∆ are block diagonal
(λΘ –i 1)∆ –i 1 = ∆ i Θ i ,
–1
(4.1)
and right coprime, from (4.4), and since PF = Θ∆–1, the set
of all block diagonal stabilizing controllers for PF is
where ∆i and ∆ i are equivalent, and λ Θ –i 1 and Θ i are
equivalent. Since det ∆ i (∞) ≠ 0 and ∆ i and ∆ i are
equivalent,∆ i is proper, 1 ≤ i ≤ k.
–1

A PF = (αΘ + ∆Q d)(βψ∆ – 1 – ΘQ d) – 1
Let ∆ i = diag{∆ i} i = 1 , and let
k
d

F = P Θ∆
–1 –1
. (4.2)
1
Since P–1Θ = DrΓ and ∆–1 are proper, F in (4.2) is a proper Q d is block diagonal and stable . (4.5)
rational matrix, and the cascade connection of F and P has
a block diagonal transfer matrix PF =: PF. Clearly, PF =
Θ∆ i –1. To show that the cascade connection of F and P is Hence,
stabilizable, we only have to show that there is a controller
A(s) such that the system S(P, F , A) shown in Fig. 2 is A = (αΘ + ∆Q d)(βψ∆ – 1 – ΘQ d) – 1 (4.6)
stable. We will do this by showing that if A(s) is a block
diagonal controller which stabilizes the plant PF , that is, for some Qd which is block diagonal and stable. Since
the system S(PF , A) shown in Fig. 3 is stable, then the S(PF, A) is stable, Hyiuj, i = 2, 3 and j = 1, 3, are stable. In
system S(P, F , A) is also stable. Let u := [u 1T u 2T u 3T ]T and the Appendix, we will show that the transfer matrices
y := [y1T y2T y3T ]T. The closed-loop transfer matrix H yu of Hy1u1, Hy1u2, Hy1u3, Hy2u2 and Hy3u2 are also stable. Hence,
the system S(P, F , A) is the system S(P, F , A) is stable and F defined in (4.2) is
C.-A. Lin and C.-M. Wu: Block Decoupling Control of Linear Multivariable Systems 151

an admissible block decoupling precompensator for P.


I 0 0
We summarize this result in the following theorem.
0 U– 1 0 . (4.9)
0 0 I
Theorem 4.2. For the plant P defined in (3.1), the block
decoupling precompensator F = P–1Θ∆–1 is admissible.
Since the transfer matrices of PF and PF are the same
Comment. It is also possible to generate the set of all and have no CI+ hidden mode, the set of all block decoupl-
admissible block decoupling precompensators for P ing controllers for PF is the same as the set of all block
from F . Let R be a block diagonal matrix with block decoupling controllers for PF . Since U and U –1 are
diagonal entry Ri ∈ IR(s)ni × ni. It is clear that if the cascade S-unimodular and the sets of the block decoupling con-
connection of Ri and Θi∆ i have no CI+ pole-zero can-
–1 trollers for PF and PF are the same, the system S(P, F, A) is
cellation and G = F R is proper, then G is also an admissible stable if and only if the system S(P, F , A) is stable and F
block decoupling precompensator for P. The converse is satisfies that F (I + PF A)–1, F AP(I + F AP)–1 and F (I +
also true: if there is CI+ pole-zero cancellation in the cas- APF )–1 are stable. Obviously, if we choose F ∈ S(m – n) × n,
then F defined in (4.7) is an admissible block decoupling
cade connection of Ri and Θi∆ i , then G = F R is not
–1
precompensator for P.
admissible. A detailed proof can be found in [10]. The I/O map Hy2u1 of S(P, F, A) is the same as the
I/O map Hy2u1 of S(P, F , A). By computation, the set of all
We have constructed an admissible block decoupl- I/O maps of S(P, F, A) is
ing precompensator for a square plant that satisfies
assumptions A1) and A2). We will now consider the {αλI + ΘQd∆|Qd is block diagonal and stable}.
rectangular plant P ∈ IRpo(s)n × m. Define (4.10)

F=U F , (4.7) Thus, the block decoupling precompensator F does


F not introduce any constraints placed on the achievable I/O
maps.
where F ∈ IRpo(s)n × n is an admissible block decoupl-
ing precompensator for P defined in (4.2) and F ∈ V. CONCLUSIONS
IR po(s)(m – n) × n. Let the transfer matrix of the cascade
connection of F and P be PF. The transfer matrix of PF is In this paper, we have described parameterizations
the same as that of PF , which is block diagonal. To show of the set of all block decoupling controllers and the set of
that the cascade connection of F and P is stabilizable, all block diagonal I/O maps. The computations involve
and hence, that F is admissible, we have to show that if A coprime factorizations and inverse of rational matrices.
is a block diagonal controller such that the system S(PF, A) Through construction of an admissible open-loop block
is stable, then the system S(P, F, A) is also stable. The decoupling precompensator, it has been shown that the
closed-loop transfer matrix of the system S(P, F, A) is block decoupling controller design reduces to a set of k
block channel designs of smaller dimensions provided that
the design requirement is based on the I/O map and the
FA(I + PFA) – 1 – FAP(I + FAP) – 1 F(I + APF) – 1 sensitivity map.
H yu = PFA(I + PFA) – 1 P(I + FAP) – 1 PF(I + APF) – 1
–1
A(I + PFA) – AP(I + FAP) – APF(I + APF) – 1
–1
APPENDIX

We will first show that λΘ i ∈ Sni × ni and ψ∆ i ∈


–1 –1

S . Thus, λΘ and ψ∆ are stable. Since λΘ –i 1 and Θ i


ni × ni –1 –1
U0 0
= 0 I 0 . (4.8) are equivalent, there exist Xi, Yi ∈ Sni × ni, S-unimodular,
0 0 I
such that, λΘ –i 1 = XiΘ i Yi. Hence, λΘ i = YiΘiXi and λΘ i
–1 –1

∈ Sni × ni, 1 ≤ i ≤ k. Thus, λΘ is stable. Similarly, ∆i and


–1

∆ i are equivalent, and ψ∆ i and ψ∆ i are also equivalent.


–1 –1

F(I + PFA) – 1 FAP(I + FAP) – 1 F(I + APF) – 1 By Proposition 4.1, ψ∆ i ∈ Sni × ni; thus, ψ∆ i ∈ Sni × ni and
0 –1 –1

F(I + PFA) – 1 FAP(I + FAP) – 1 0 F(I + APF) – 1
ψ∆ is stable.
–1

By (4.1), (4.2) and (4.6),


PFA(I + PFA) – 1 P(I + FAP) – 1 0 PF(I + APF) – 1

–1

= I + Θ∆ (αΘ + ∆Q d)(βψ∆ – ΘQ d)
–1 –1 –1 –1
A(I + PFA) –1
– AP(I + FAP) –1
0 – APF(I + APF) –1
(I + PFA)
(A.1)
152 Asian Journal of Control, Vol. 1, No. 3, September 1999

= βψI – ΘQ d∆ . (A.2) Vol. 27, No. 1, pp. 47-53 (1996).


6. Lin, C.A. and C.M. Wu, “Block-Decoupling Linear
By (3.16), Multivariable Systems: Necessary and Sufficient
Conditions,” Automatica, Vol. 34, No. 2, pp. 237-243
–1 (1998).
H y1u1 = FA(I + PFA) – 1 = P I – (I + PFA) – 1 (A.3)
7. Linnemann, A. and Q.G. Wang, “Block Decoupling
with Stability by Unity Output Feedback —Solution
= D rΓ(αλΘ – 1 + Q d∆) . (A.4) and Performance Limitations,” Automatica, Vol. 29,
No. 3, pp. 735-744 (1993).
By (3.16) and (3.19), 8 Nett, C.N., C.A. Jacobson and M.J. Balas, “A Connec-
–1
tion between State-space and Doubly Coprime Frac-
H y1u2 = – FAP(I + FAP) – 1 = – I + P (I + PFA) – 1P tional Representation,” IEEE Trans. Automat. Contr.,
(A.5) Vol. 29, pp. 831-832 (1984).
9. Vidyasagar, M., Control System Synthesis: A Factor-
= – αλI – D rΓQ dΩN l . (A.6) ization Approach, MIT Press, MA (1985).
10 Wu, C.M., “Block Decoupling Control of Linear
Furthermore, Multivariable Feedback Systems,” Ph.D. Dissertation,
Department of Electrical and Control Engineering,
H y1u3 = F(I + APF) – 1 = P(I + PFA) – 1PF (A.7) National Chiao-Tung University, Hsinchu, Taiwan
(2000).
= βD rΓψ∆ – D rΓQ dλΘ
–1 –1
; (A.8)

H y2u2 = P(I + FAP) – 1 = (I + PFA) – 1P (A.9)


Ching-An Lin received the B.S. de-
= (βψ∆ – ΘQ d)ΩN l ;
–1
(A.10) gree from National Chiao-Tung
University, Taiwan, in 1977, the M.S.
H y3u2 = – AP(I + FAP) – 1 (A.11) degree from the University of New
Mexico, Albuquerque, in 1980, and
= – (αΘ + ∆Q d)ΩN l . (A.12) the Ph.D. degree from the University
of California, Berkeley, 1984, all in
Since λΘ–1, λΘ , ψ∆–1 and ψ∆ are stable, the
–1 –1
electrical engineering.
transfer matrices Hy1u1, Hy1u2, Hy1u3, Hy2u2 and Hy3u2 are He was with the Chung-Shan Institute of Science and
stable. Technology from 1977 to 1979, and with Integrated Sys-
tems Inc. from 1984 to 1986. Since June 1986, he has been
REFERENCES with the Department of Electrical and Control Engineering,
National Chiao-Tung University, Taiwan, where he is a
1. Anderson, B.D.O. and M.R. Gevers, “On Multivari- professor. His current research interests are in multivari-
able Pole-Zero Cancellations and the Stability of Feed- able control and multirate signal processing.
back Systems,” IEEE Trans. Circuits Syst., Vol. 28,
pp. 830-833 (1981).
2. Callier, F.M. and C.A. Desoer, Multivariable Feed-
back Systems, Springer-Verlag, New York (1982). Chang-Ming Wu was born in
3. Commault, C., J.M. Dion and J.A. Torres, “Minimal Taiwan, Taiwan, R.O.C., on May 27,
Structure in the Block Decoupling Problem with 1969. He received the B.S. degree in
Stability,” Automatica., Vol. 27, No. 2, pp. 331-338 control engineering from National
(1991). Chiao-Tung University, Hsinchu,
4. Desoer, C.A. and A.N. Gundes, “Decoupling Linear Taiwan, in 1991.
Multiinput Multioutput Plants by Dynamic Output He is currently working toward
Feedback: An Algebraic Theory,” IEEE Trans. the Ph.D. degree in electrical and con-
Automat. Contr., Vol. AC-31, No. 8, pp. 744-750 trol engineering at National Chiao-Tung University, Taiwan.
(1986) His research interests include multivariable control sys-
5. Lin, C.A., C.W. Yang and T.F. Hsieh, “An Algorithm tems and multirate systems.
for Inverting Rational Matrices,” Syst. Contr. Lett.,

Anda mungkin juga menyukai