Professional Documents
Culture Documents
Driggers Presentation 031606
Driggers Presentation 031606
Control Designs
Dennis Driggers
EE691
March 16, 2006
Overview
Most types of system control techniques
incorporate some type of disturbance
waveform modeling. Even if the
disturbance waveform is completely
unknown, a disturbance characterization
of the waveform is assumed. This
assumption is usually made on a worst
case basis to insure stability of the
targeted system.
Classical and Modern Control theory
incorporates waveform characterization of
disturbances with and without waveform
structure. Modern control theory is
centered around modeling the disturbance
to either completely reject, minimize, or to
even utilize the disturbance in controlling
system behavior. In all of these
circumstances it is necessary to model the
waveform.
Some Waveform Models used in
Modern Control Design
) ( ) ( ) (
) cos( ) sin( ) cos( ) sin( ) (
) ( ) (
) ( ) (
) ( ) (
2
3
3
2 2
5
5
5 4 3 2 1
2
2
2 1
2
2
2 1
t
dt
dw
dt
w d
dt
w d
t c t c t c t c c t w
t
dt
dw
dt
w d
e c c t w
t
dt
w d
t c c t w
t
dt
dw
c t w
t
e u| | u
| | u u
e o
e
o
o
= + + +
+ + + + =
= + + =
= + =
= =
Introduction to Sliding Mode
Control
Sliding Mode Control does not require a disturbance
waveform characterization to implement the control law.
The main advantage of Sliding Mode Control (SMC) is
the robustness to unknown disturbances. Required
knowledge of the disturbance is limited to the
disturbance boundary. Traditional SMC was, however,
limited by a discontinuous control law. Depending on the
plant dynamics, high frequency switching may or may
not be an issue to contend with. There are techniques to
limit and eliminate the high-frequency switching
associated with traditional SMC. It is the intent of this
paper to look at several SMC techniques utilizing an
aircraft model with bounded external disturbances.
Agenda
Background of SMC
Definitions
SMC Design Methodology
Derivations
Traditional SMC
Supertwist
SMC driven by SMC observer
Simulation Results
Conclusions
Agenda
Background of SMC
Definitions
SMC Design Methodology
Derivations
Traditional SMC
Supertwist
SMC driven by SMC observer
Simulation Results
Conclusions
Background
Sliding Mode Control (SMC) theory was founded and
advanced in the former Soviet Union as a variable
structure control system.
SMC is a relatively young control concept dating back to
the 1960s.
SMC theory first appeared outside Russia in the mid
1970s when a book by Itkis (1976) and a survey paper
by Utkin (1977) were published in English.
The SMC reachability condition is based on the
Russian mathematician, Lyapunov, and his theory of
stability of nonlinear systems.
Agenda
Background of SMC
Definitions
SMC Design Methodology
Derivations
Traditional SMC
Supertwist
SMC driven by SMC observer
Simulation Results
Conclusions
Definitions
State Space An n-dimensional space whose
coordinate axes consist of the x1 axes,x2
axis,,x
n
axes.
State trajectory- A graph of x(t) verses t through
a state space.
State variables The state variables of a system
consist of a minimum set of parameters that
completely summarize the systems status.
Disturbance Completely or partially unknown
system inputs which cannot be manipulated by
the system designer.
Definitions
Sliding Surface A line or hyperplane in
state-space which is designed to
accommodate a sliding motion.
Sliding Mode The behavior of a dynamical
system while confined to the sliding surface.
Signum function (Sign(s))
Reaching phase The initial phase of the
closed loop behaviour of the state variables
as they are being driven towards the
surface.
<
> +
0 ) , ( 1
0 ) , ( 1
y y if s
y y s if
Agenda
Background of SMC
Definitions
SMC design Methodology
Derivations
Traditional SMC
Supertwist
SMC driven by SMC observer
Simulation Results
Conclusions
SMC Design Methodology
Three Basic Steps
Design a sliding manifold or sliding surface
in state space.
Design a controller to reach the sliding
surface in finite time.
Design a control law to confine the desired
state variables to the sliding manifold.
SMC Graphical Illustration
Agenda
Background of SMC
Definitions
SMC design Methodology
Derivations
Traditional SMC
Supertwist
SMC driven by SMC observer
Simulation Results
Conclusions
Aircraft Modeled Parameters
Simplified aircraft model consist of angle of attack,
aircraft pitch rate, and elevator deflection represented
as ,q, and
e.
Aircraft parameters for a particular airframe at a
particular attitude and altitude.
Changes in airframe due to damage (unknown,
uncertain, and bounded)
Horizontal tail and rudder areas.
Flight profile filters.
A A
n
A
B
* o
o
c
Aircraft and Disturbance Models
used in Simulations
u
e e
20 20 + = o o
,
~
e
B
q
A
q
o
o o
+
(
=
(
,
~
A A A
n
A + =
(
=
26 . 1 72 . 3
99 . 15 . 1
An
), 3 (
42 . 0 85 . 1
003 . 0 04 . 0
(
= A t U A
,
5 . 19
0
(
= B
where
). (
4 3
4
2
t U
s s
c
+ +
=
-
o
o
) 3 sin( ) ( t t =
0 ) 0 ( , 1 . 0 ) 0 ( , 0 ) 0 (
:
= = = q and
Conditions Initial
c
o o
Derivations for Traditional SMC
It is necessary to find the relative degree of the system in
state-space. Relative degree, , is determined by the
number of times the output has to be differentiated before
any control input appears in its expression.
The aircraft model in scalar format is:
The relative degree of the plant is 3 as the control u
appears as follows:
e
q q
q
o o
o o
5 . 19 26 . 1 72 . 3
99 . 0 15 . 1
=
+ =
I
bu h gq f y
e
+ + + = = o o o
o = y
u e e 20 20 + = o o
o
o o
o o
o
o
o o o o
o
o
) 0 (
: by (0) any for satisfied is
condition ty reachabili the zero to setting By
) 0 (
) 0 (
) 0 ( ) 0 ( ) (
dt
0
) (
) 0 (
s
= =
=
= =
s s
} }
r
r
t t
t
t t
t t
dt d
d
then
constant. positive small a is o o o where s
Reaching Phase Design
Introduce a Lyapunov function candidate.
The derivative of the Lyapunov function is
The initial conditions are given as:
and .
Desire seconds, then
04 . 17
5 . 0
52 . 8 ) 0 ( ) 0 (
=
= > s
r
r
t
t
o
o
o o o s =
V
2
2
1
) ( o o = V
, 1 . 0 ) 0 ( = o 0 ) 0 ( = q , 0 ) 0 ( =
c
o
8.52 100(0.1) 1.15(0.1)) 14( 1.32(0.1) (0) = =
5 . 0 =
r
t
) ( 100 ) ( 14 o o o o o o o + + =
c c c
SMC Controller Design
The controller can be implemented with the
signum function as follows:
( ) ) (o sign L u + =
) 3 sin( t L >
. 5 . 1 let 1 ~ > L L
04 . 17
5 . 0
52 . 8
= >
) ( 5 . 18 o sign u =
Simulink Diagram for Traditionial
SMC
Supertwist Design
It has been shown (not in this brief) that the solution
to the following differential equation
and its derivative converge to zero in finite time if
, , and .
On this basis u is introduced as:
o o | o o o d sign sign u ) ( ) (
2 / 1
}
+ =
L 2 / 1 > o
L 4 > |
L t s ) (
) ( ) ( ) (
2 / 1
t d z sign z sign z z t | o = + +
}
Supertwist Design
Supertwist utilizes the same sliding surface and
values as the traditional SMC. The signum
control function is replaced with the function:
The values for L=1.5 are:
o o | o o o d sign sign u ) ( ) (
2 / 1
}
+ =
L 2 / 1 > o L 4 > |
612 . 5 . 1 2 / 1 = > o
, 6 ) 5 . 1 ( 4 = > |
Supertwist Block Diagram
SMC Observer Design
. ) (
) 3 ( )) ( ( ) ) ( (
: yields ) (
with ng substituti and Eq(2) of inequality the to (1) Eq. Applying
(2) 0
(1) ) (
) (
: yields solving and into ng substituti and ating Differenti
: follows as
designed is and as introduced is variable sliding auxiliary new A
0
1 1
o
o
+ > s
+ s + =
+
> <
+ =
+ + = + =
+ = + =
= > + =
+ = = + =
L and L where
s wsign L s w s s s
s wsign L
s s s
w s
w v v z s
z s
w v z and z s
s
) ( and v , K v -K v
v where ) ( v, bu, bu ) (
SMC Observer Design
). (
w w
w
w s
s s w
s )sign(s)) (L s(
s sign )sign(s) L-(L s )sign(s) -(L s
)sign(s) -(L w
w
eq