beautypg.com

Hankel singular values, Hankel singular values -8 – National Instruments NI MATRIXx Xmath User Manual

Page 15

background image

Chapter 1

Introduction

Xmath Model Reduction Module

1-8

ni.com

The controllability grammian is also E[x(t)x

′(t)] when the system

has been excited from time –

∞ by zero mean white

noise with

.

The observability grammian can be thought of as measuring the
information contained in the output concerning an initial state.
If

with

x(0) = x

0

then:

Systems that are easy to observe correspond to Q with large
eigenvalues, and thus large output energy (when unforced).

lyapunov(A,B*B')

produces P and

lyapunov(A',C'*C)

produces Q.

For a discrete-time G(z) = D + C(zI-A)

–1

B with |

λ

i

(A)|<1, P and Q are:

P – APA

′ = BB′

Q – A

′QA = C′C

The first dot point above remains valid. Also,

and

with the sums being finite in case A is nilpotent (which is the case if
the transfer-function matrix has a finite impulse response).

[IA

A] vec P = vec (BB′)

lyapunov( )

can be used to evaluate P and Q.

Hankel Singular Values

If P, Q are the controllability and observability grammians of a
transfer-function matrix (in continuous or discrete time), the Hankel
Singular Values
are the quantities

λ

i

1/2

(PQ). Notice the following:

All eigenvalues of PQ are nonnegative, and so are the Hankel singular
values.

The Hankel singular values are independent of the realization used to
calculate them: when A,B,C,D are replaced by TAT

–1

, TB, CT

–1

and D,

then P and Q are replaced by TPT

′ and (T

–1

)

QT

–1

; then PQ is replaced

by TPQT

–1

and the eigenvalues are unaltered.

The number of nonzero Hankel singular values is the order or
McMillan degree of the transfer-function matrix, or the state
dimension in a minimal realization.

x·

Ax Bw

+

=

E w t

( )ws

( )

[

]

I

δ t s

(

)

=

x·

Ax

=

y

,

Cx

=

y

t

( )y t

( )dt

0

x

0

Qx

0

=

P

A

k

BB

A

k

k

0

=

=

Q

A

k

C

CA

k

k

0

=

=