RICCARDO SCATTOLINI – ADVANCED AND MULTIVARIABLE CONTROL - NORMS

Advanced and Multivariable Control

Norms, gains, small gain theorem

Riccardo Scattolini

1
RICCARDO SCATTOLINI – ADVANCED AND MULTIVARIABLE CONTROL – NORMS, GAIN, STABILITY OF FEEDBACK SYSTEMS

Norms of vectors

of numbers
2 norm

inf norm

of signals
RICCARDO SCATTOLINI – ADVANCED AND MULTIVARIABLE CONTROL – NORMS, GAIN, STABILITY OF FEEDBACK SYSTEMS

Singular values
RICCARDO SCATTOLINI – ADVANCED AND MULTIVARIABLE CONTROL – NORMS, GAIN, STABILITY OF FEEDBACK SYSTEMS

Singular value decomposition

Unitary matrix
RICCARDO SCATTOLINI – ADVANCED AND MULTIVARIABLE CONTROL – NORMS, GAIN, STABILITY OF FEEDBACK SYSTEMS

Minimum and maximum singular values

RICCARDO SCATTOLINI – ADVANCED AND MULTIVARIABLE CONTROL – NORMS, GAIN, STABILITY OF FEEDBACK SYSTEMS

Induced p-norm of a matrix

Induced 2-norm

Norm of a «map» A

u y
A

gain of the map

RICCARDO SCATTOLINI – ADVANCED AND MULTIVARIABLE CONTROL – NORMS, GAIN, STABILITY OF FEEDBACK SYSTEMS

Norm of systems

u(t) y(t)
S(u(t)) not necessarily linear

Gain γ of S
RICCARDO SCATTOLINI – ADVANCED AND MULTIVARIABLE CONTROL – NORMS, GAIN, STABILITY OF FEEDBACK SYSTEMS

Example of linear system – the integrator

Input u = sca (t)

Bode Diagram
20

10

Magnitude (dB)
0

-10

infinite gain -20

-89

-89.5

Phase (deg)
-90

-90.5

-91
-1 1
10 10 0 10
Frequency (rad/s)
RICCARDO SCATTOLINI – ADVANCED AND MULTIVARIABLE CONTROL – NORMS, GAIN, STABILITY OF FEEDBACK SYSTEMS

Example – SISO asymptotically stable linear system

Parseval theorem

=
RICCARDO SCATTOLINI – ADVANCED AND MULTIVARIABLE CONTROL – NORMS, GAIN, STABILITY OF FEEDBACK SYSTEMS

Extension to MIMO asymptotically stable linear system

Singular Values
70

60

50

Singular Values (dB)

40

30

20

10
-4 -3 -2 -1
10 10 10 10 10 0
Frequency (rad/s)

Input – Output (I/O) stability

RICCARDO SCATTOLINI – ADVANCED AND MULTIVARIABLE CONTROL – NORMS, GAIN, STABILITY OF FEEDBACK SYSTEMS

Singular Values
50

40

30

20

Singular Values (dB)

10

-10

g11=tf(100,[1 1]); -20

g12=tf(10,conv([1 1],[1 2])); -30

-2 -1 0 1 2

g21=tf(10,[1 2]);
10 10 10 10 10
Frequency (rad/s)

g22=tf(10,[1 2]);
G1=[g11 g12;g21 g22]
sigma(G1)
RICCARDO SCATTOLINI – ADVANCED AND MULTIVARIABLE CONTROL – NORMS, GAIN, STABILITY OF FEEDBACK SYSTEMS

Different definitions of gain for asymptotically stable linear systems

It depends on the applied input

RICCARDO SCATTOLINI – ADVANCED AND MULTIVARIABLE CONTROL – NORMS, GAIN, STABILITY OF FEEDBACK SYSTEMS

2-norm gain for asymptotically stable, strictly proper, linear systems

SISO

MIMO
RICCARDO SCATTOLINI – ADVANCED AND MULTIVARIABLE CONTROL – NORMS, GAIN, STABILITY OF FEEDBACK SYSTEMS
RICCARDO SCATTOLINI – ADVANCED AND MULTIVARIABLE CONTROL – NORMS, GAIN, STABILITY OF FEEDBACK SYSTEMS
RICCARDO SCATTOLINI – ADVANCED AND MULTIVARIABLE CONTROL – NORMS, GAIN, STABILITY OF FEEDBACK SYSTEMS
RICCARDO SCATTOLINI – ADVANCED AND MULTIVARIABLE CONTROL – NORMS, GAIN, STABILITY OF FEEDBACK SYSTEMS

Small gain theorem (one of the most useful tools for the analysis of nonlinear feedback systems)

less restrictive

only sufficient conditions

RICCARDO SCATTOLINI – ADVANCED AND MULTIVARIABLE CONTROL – NORMS, GAIN, STABILITY OF FEEDBACK SYSTEMS

Proof
RICCARDO SCATTOLINI – ADVANCED AND MULTIVARIABLE CONTROL – NORMS, GAIN, STABILITY OF FEEDBACK SYSTEMS

SISO Linear systems

Nyquist criterion:

N=0
P=0

S1(jω)S2(jω)
RICCARDO SCATTOLINI – ADVANCED AND MULTIVARIABLE CONTROL – NORMS, GAIN, STABILITY OF FEEDBACK SYSTEMS

Stability of feedback systems with static sector nonlinearity

asymptotically stable SISO system

In view of the small gain theorem, I/O

stability of the feedback system is
guaranteed if
RICCARDO SCATTOLINI – ADVANCED AND MULTIVARIABLE CONTROL – NORMS, GAIN, STABILITY OF FEEDBACK SYSTEMS

A less stringent condition : THE CIRCLE CRITERION (proof in the textbook)

RICCARDO SCATTOLINI – ADVANCED AND MULTIVARIABLE CONTROL – NORMS, GAIN, STABILITY OF FEEDBACK SYSTEMS
RICCARDO SCATTOLINI – ADVANCED AND MULTIVARIABLE CONTROL – NORMS, GAIN, STABILITY OF FEEDBACK SYSTEMS

THE CIRCLE CRITERION comments and interpretations

Only a sufficient condition

Can be generalized to non asymptotically stable G(s)

When k10 the circle becomes a vertical line passing through -1/k2

Interesting cases, widely used in practical applications

saturation dead zone

RICCARDO SCATTOLINI – ADVANCED AND MULTIVARIABLE CONTROL – NORMS, GAIN, STABILITY OF FEEDBACK SYSTEMS

EXAM June 2019

solution
RICCARDO SCATTOLINI – ADVANCED AND MULTIVARIABLE CONTROL – NORMS, GAIN, STABILITY OF FEEDBACK SYSTEMS