Robust Adaptive Control by Petros A. Ioannou, Jing Sun - HTML preview

PLEASE NOTE: This is an HTML preview only and some elements such as links or page numbers may be incorrect.
Download the book in PDF, ePub, Kindle for a complete version.

control laws that can meet the pole placement control objective when the

plant parameters are known exactly. The form of these control laws as well

as the mapping between the controller and plant parameters will be used

in Section 7.4 to form indirect APPC schemes for plants with unknown

parameters.

7.3.1

Problem Statement

Consider the SISO LTI plant

Z

y

p( s)

p = Gp( s) up,

Gp( s) =

(7.3.1)

Rp( s)

where Gp( s) is proper and Rp( s) is a monic polynomial. The control objec-

tive is to choose the plant input up so that the closed-loop poles are assigned

to those of a given monic Hurwitz polynomial A∗( s). The polynomial A∗( s),

referred to as the desired closed-loop characteristic polynomial, is chosen

based on the closed-loop performance requirements. To meet the control

objective, we make the following assumptions about the plant:

P1. Rp( s) is a monic polynomial whose degree n is known.

P2. Zp( s) , Rp( s) are coprime and degree( Zp) < n.

Assumptions (P1) and (P2) allow Zp, Rp to be non-Hurwitz in contrast to the

MRC case where Zp is required to be Hurwitz. If, however, Zp is Hurwitz,

the MRC problem is a special case of the general pole placement problem

defined above with A∗( s) restricted to have Zp as a factor. We will explain

the connection between the MRC and the PPC problems in Section 7.3.2.

In general, by assigning the closed-loop poles to those of A∗( s), we can

guarantee closed-loop stability and convergence of the plant output yp to zero

provided there is no external input. We can also extend the PPC objective

to include tracking, where yp is required to follow a certain class of reference

signals ym, by using the internal model principle discussed in Chapter 3 as

follows: The reference signal ym ∈ L∞ is assumed to satisfy

Qm( s) ym = 0

(7.3.2)

7.3. PPC: KNOWN PLANT PARAMETERS

449

where Qm( s), the internal model of ym, is a known monic polynomial of

degree q with nonrepeated roots on the -axis and satisfies

P3. Qm( s) , Zp( s) are coprime.

For example, if yp is required to track the reference signal ym = 2 + sin(2 t),

then Qm( s) = s( s 2 + 4) and, therefore, according to P3, Zp( s) should not

have s or s 2 + 4 as a factor.

The effect of Qm( s) on the tracking error e 1 = yp − ym is explained

in Chapter 3 for a general feedback system and it is analyzed again in the

sections to follow.

In addition to assumptions P1 to P3, let us also assume that the co-

efficients of Zp( s) , Rp( s), i.e., the plant parameters are known exactly and

propose several control laws that meet the control objective. The knowledge

of the plant parameters is relaxed in Section 7.4.

7.3.2

Polynomial Approach

We consider the control law

Qm( s) L( s) up = −P ( s) yp + M( s) ym

(7.3.3)

where P ( s) , L( s) , M ( s) are polynomials (with L( s) monic) of degree q + n −

1 , n − 1 , q + n − 1, respectively, to be found and Qm( s) satisfies (7.3.2) and

assumption P3.

Applying (7.3.3) to the plant (7.3.1), we obtain the closed-loop plant

equation

Z

y

pM

p =

y

LQ

m

(7.3.4)

mRp + P Zp

whose characteristic equation

LQmRp + P Zp = 0

(7.3.5)

has order 2 n + q − 1. The objective now is to choose P, L such that

LQmRp + P Zp = A∗

(7.3.6)

is satisfied for a given monic Hurwitz polynomial A∗( s) of degree 2 n + q − 1.

Because assumptions P2 and P3 guarantee that QmRp, Zp are coprime, it

450

CHAPTER 7. ADAPTIVE POLE PLACEMENT CONTROL

follows from Theorem 2.3.1 that L, P satisfying (7.3.6) exist and are unique.

The solution for the coefficients of L( s) , P ( s) of equation (7.3.6) may be

obtained by solving the algebraic equation

Slβl = α∗l

(7.3.7)

where Sl is the Sylvester matrix of QmRp, Zp of dimension 2( n+ q) × 2( n+ q) βl = [ lq , p ] , a∗l = [0 , . . . , 0 , 1 , α∗ ]

q

lq = [0 , . . . , 0 , 1 , l ] ∈ Rn+ q

q

l = [ ln− 2 , ln− 3 , . . . , l 1 , l 0] ∈ Rn− 1

p = [ pn+ q− 1 , pn+ q− 2 , . . . , p 1 , p 0] ∈ Rn+ q

α∗ = [ a∗ 2 n+ q− 2 , a∗ 2 n+ q− 3 , . . . , a∗ 1 , a∗ 0] ∈ R 2 n+ q− 1

li, pi, a∗i are the coefficients of

L( s) = sn− 1 + ln− 2 sn− 2 + · · · + l 1 s + l 0 = sn− 1 + l αn− 2( s) P ( s) = pn+ q− 1 sn+ q− 1 + pn+ q− 2 sn+ q− 2 + · · · + p 1 s + p 0 = p αn+ q− 1( s) A∗( s) = s 2 n+ q− 1+ a∗ 2 n+ q− 2 s 2 n+ q− 2+ · · ·+ a∗ 1 s+ a∗ 0 = s 2 n+ q− 1+ α∗ α 2 n+ q− 2( s) The coprimeness of QmRp, Zp guarantees that Sl is nonsingular; therefore,

the coefficients of L( s) , P ( s) may be computed from the equation

βl = S− 1 α∗

l

l

Using (7.3.6), the closed-loop plant is described by

Z

y

pM

p =

y

A∗

m

(7.3.8)

Similarly, from the plant equation in (7.3.1) and the control law in (7.3.3)

and (7.3.6), we obtain

R

u

pM

p =

y

A∗

m

(7.3.9)

7.3. PPC: KNOWN PLANT PARAMETERS

451

ym

−e 1

P ( s)

u

✲ ❧

p

yp

+ Σ

G

Q

p( s)

m( s) L( s)

Figure 7.2 Block diagram of pole placement control.

Because ym ∈ L∞ and ZpM

A∗ , RpM

A∗

are proper with stable poles, it follows

that yp, up ∈ L∞ for any polynomial M( s) of degree n + q − 1. Therefore,

the pole placement objective is achieved by the control law (7.3.3) without

having to put any additional restrictions on M ( s) , Qm( s). When ym = 0,

(7.3.8), (7.3.9) imply that yp, up converge to zero exponentially fast.

When ym = 0, the tracking error e 1 = yp − ym is given by

Z

Z

LR

e

pM − A∗

p

p

1 =

y

( M − P ) y

Q

A∗

m = A∗

m − A∗ mym

(7.3.10)

For zero tracking error, (7.3.10) suggests the choice of M ( s) = P ( s) to null

the first term in (7.3.10). The second term in (7.3.10) is nulled by using

Qmym = 0. Therefore, for M( s) = P ( s), we have

Z

LR

e

p

p

1 =

[0]

[0]

A∗

A∗

Because Zp

A∗ , LRp

A∗

are proper with stable poles, it follows that e 1 converges

exponentially to zero. Therefore, the pole placement and tracking objective

are achieved by using the control law

QmLup = −P ( yp − ym)

(7.3.11)

which is implemented as shown in Figure 7.2 using n + q − 1 integrators

to realize C( s) =

P ( s)

. Because L( s) is not necessarily Hurwitz, the

Qm( s) L( s)

realization of (7.3.11) with n+ q − 1 integrators may have a transfer function,

namely C( s), with poles outside C−. An alternative realization of (7.3.11) is

obtained by rewriting (7.3.11) as

Λ − LQ

P

u

m

p =

u

( y

Λ

p − Λ p − ym)

(7.3.12)

452

CHAPTER 7. ADAPTIVE POLE PLACEMENT CONTROL

−e

u

1

P ( s)

y

p

yp

m

+ ❧

Σ ✲

+ ❧

Σ

G

Λ( s)

p( s)

+

Λ( s) − Qm( s) L( s) ✛

Λ( s)

Figure 7.3 An alternative realization of the pole placement control.

where Λ is any monic Hurwitz polynomial of degree n+ q− 1. The control law

(7.3.12) is implemented as shown in Figure 7.3 using 2( n + q − 1) integrators

to realize the proper stable transfer functions Λ −LQm , P . We summarize the

Λ

Λ

main equations of the control law in Table 7.1.

Remark 7.3.1 The MRC law of Section 6.3.2 shown in Figure 6.1 is a

special case of the general PPC law (7.3.3), (7.3.6). We can obtain the

MRC law of Section 6.3.2 by choosing

R

Q

mΛ0

m = 1 ,

A∗ = ZpΛ0 Rm,

M ( s) =

kp

L( s) = Λ( s) − θ∗ 1 αn− 2( s) , P ( s) = ( θ∗ 2 αn− 2( s) + θ∗ 3Λ( s)) Z

Λ = Λ

m

0 Zm,

ym = km

r

Rm

where Zp, Λ0 , Rm are Hurwitz and Λ0 , Rm, kp, θ∗ 1 , θ∗ 2 , θ∗ 3 , r are as defined in Section 6.3.2.

Example 7.3.1 Consider the plant

b

yp =

u

s + a p

where a and b are known constants. The control objective is to choose up such that

the poles of the closed-loop plant are placed at the roots of A∗( s) = ( s + 1)2 and

yp tracks the constant reference signal ym = 1. Clearly the internal model of ym is

Qm( s) = s, i.e., q = 1. Because n = 1, the polynomials L, P, Λ are of the form

L( s) = 1 ,

P ( s) = p 1 s + p 0 ,

Λ = s + λ 0

7.3. PPC: KNOWN PLANT PARAMETERS

453

Table 7.1 PPC law: polynomial approach

Plant

yp = Zp( s) u

R

p

p( s)

Reference input

Qm( s) ym = 0

Solve

for

L( s)

=

sn− 1 + l αn− 2( s) and

P ( s) = p αn+ q− 1( s) the polynomial equation

L( s) Qm( s) Rp( s) + P ( s) Zp( s) = A∗( s) Calculation

or solve for βl the algebraic equation Slβl = α∗l,

where Sl is the Sylvester matrix of RpQm, Zp

βl = [ lq , p ] ∈ R 2( n+ q)

lq = [0 , . . . , 0 , 1 , l ] ∈ Rn+ q

q

A∗( s) = s 2 n+ q− 1 + α∗ α 2 n+ q− 2( s)

α∗l = [0 , . . . , 0 , 1 , α∗ ] ∈ R 2( n+ q)

q

u

u

e

Control law

p = Λ −LQm

Λ

p − P

Λ 1

e 1 = yp − ym

A∗( s) is monic Hurwitz; Qm( s) is a monic poly-

nomial of degree q with nonrepeated roots on

Design variables

axis; Λ( s) is a monic Hurwitz polynomial of degree

n + q − 1

where λ 0 > 0 is arbitrary and p 0 , p 1 are calculated by solving

s( s + a) + ( p 1 s + p 0) b = ( s + 1)2

(7.3.13)

Equating the coefficients of the powers of s in (7.3.13), we obtain

2 − a

1

p 1 =

,

p

b

0 = b

454

CHAPTER 7. ADAPTIVE POLE PLACEMENT CONTROL

Equation (7.3.13) may be also written in the form of the algebraic equation (7.3.7),

i.e., the Sylvester matrix of s( s + a) , b is given by

1 0 0 0

a 1 0 0 

S

l =  0 a b 0 

0 0 0 b

and

0

0

 1 

 1 

β

l =  p , α∗l = 

1

2

p 0

1

Therefore, the PPC law is given by

( s + λ

2 − a

1

1

u

0 − s)

p

=

u

s +

e

s + λ

p −

1

0

b

b s + λ 0

λ

(2 − a) s + 1

=

0

u

e