Adaptive Control 2

ADAPTIVE CONTROL
SYSTEMS
MRAC
MODEL REFERENCE ADAPTIVE CONTROL
SYSTEM
It can be considered as an ADAPTIVE SERVO
SYSTEM.
It consists of two loops inner loop &outer loop
It consists of a reference model in outer loop.
Ordinary feed back is called inner loop.
Parameters are adjusted on basis of feedback
from the error.
Error is the difference between produced output
& reference model value.
Adjustment of system parameters

in a MRAC can be obtained in two
ways.
GRADIENT METHOD (MIT RULE)
LYPNOV STABILITY THEORY
MODEL
CONTROLLER
YM
PARAMETERS
ADJUSTMENT
MECHANISM
UC
U
CONTROLLER
BLOCK
DIAGRAM OF MRAC
Y
PLANT
MRAC IS COMPOSED OF
Plant containing unknown parameters
Reference model
Adjustable parameters containing control
law or loss law
Ordinary feed back loop
MIT RULE
It is developed by INSTRUMANTATION LABORATORY
at MIT.
To consider MIT rule we use a closed loop response in
which controller has one adjustable parameter .
The desired closed loop response is specified by a
model whose output is ym.
Then error e= ym y. (y is original output).
Error criteria selected here is
J()=1/2 (e)2
And this loss function is to be minimized to make the system controlled.
To achieve this the parameters are changed in the direction of negative
gradient of J.
d = - J = - e e
dt
This is called
GRADIENT or MIT rule.
=adaptation gain
e/ = sensitivity derivative(informs how error is
influenced by
The loss function can also be chosen as
J()=|e|
d = - e sign e [GRADIENT METHOD]
dt
first MRAC is implemented by this function

d = -
sign e sign (e)

dt
[ sign-sign algorithm]
used in telecommunication where simple
implementation & fast computing are required
For multivariable systems

is considered as vector
e/ is considered as gradient of the
error with respect to the parameter
APPLICATIONS FOR MIT

RULE
ADAPTATION OF A FEED FORWARD GAIN
MRAS FOR A FIRST ORDER SYSTEM
ADAPTATION OF A FEED FORWARD

GAIN
PROBLEM :Adjustment of feed forward gain
ASSUMPTIONS:
Process is linear with the transfer function KG(s)
G(s) is known
K is unknown parameter
DESIRED CONDITION:
Transfer function Gm(s) should be equal to KoG(s)
Ko is given constant
MODEL
Ym
KOG(S)
-/S
UC
PROCESS
KG(S)
MRAC FOR ADJUSTMENT OF FEEDFORWARD GAIN BY MIT RULE
The constant k for this plant is unknown. However, a

reference model can be formed with a desired value of k,
and through adaptation of a feedforward gain, the
response of the plant can be made to match this model.
The reference model is therefore chosen as the plant
multiplied by a desired constant ko .
The cost function chosen here is,
The error is then restated in terms of the transfer functions

multiplied by their inputs.
As can be seen, this expression for the error

contains the parameter theta which is to be updated.
To determine the update rule, the sensitivity
derivative is calculated and restated in terms of the
model ouput:
Finally, the MIT rule is applied to give an expression for

updating theta. The constants k and ko are combined into
gamma.
To tune this system, the values of ko and gamma can be

varied.
NOTES ON DESIGN WITH MIT

It is important to note that the MIT rule by itself does not
guarantee convergence or stability.
An MRAC designed using the MIT rule is very sensitive
to the amplitudes of the signals.
As a general rule, the value of gamma is kept small.
Tuning of gamma is crucial to the adaptation rate and
stability of the controller.
IF G(S)=1/S+1
UC is sinusoidal signal with frequency 1
rad/s
K=1
K0=2
Vary values of
=0.5,1,2
We can observe the output approaches the
model output at =1
simulation results
OBSERVATIONS FROM GRAPH:

Convergence rate depends on adaptation gain .
Reasonable value of has to be selected .
Applications:
In robots with unknown load.
CD player where sensitivity of laser diode is not
known.
MRAC FOR A FIRST ORDER

SYSTEM
Process
dy = ay+ bu
dt
Model
dym = amym+bmuc
dt
Controller
U(t)=1Uc(t)- 2 Y(t)
1 & 2 are 2 adjustable parameters
LYAPUNOV THEOREM
ALGORITHM FOR LYAPUNOV THEORY
Derive differential equation for the error e=y-ym.
We attempt to find lyapunov function & an adaptation
mechanism such that the error will go to zero.
We find dv/dt is usually only negative semi definite .so
finding error equation & lyapunov function with a
bounded second derivative .
It is to show bounded ness & that error goes to zero.
To show parameter convergence ,it is necessary to
impose further conditions ,such as persistently excitation
and uniform observability
MODEL
KOG(S)
YM
UC
-/S
PROCESS
KG(S)
B. DIAG FOR FEED FORWARD GAIN CONTROL USING LYAPUNOV RULE
The error equations are
The value of dv/dt is <=0 [which is ve semi

definite]
As time derivative of Lypanov function is negative
semi definite.
By using lemma of lypanov we can show error
goes to zero.
http://www.control.hut.fi/Kurssit/AS-74.185/luennot/lu5ep.pdf
http://www.control.hut.fi/Kurssit/AS-74.185/luennot/lu6ep.pdf
http://www.control.lth.se/~FRT050/Exercises/ex4sol.pdf
http://www.control.lth.se/%7EFRT050/Exercises/ex5sol.pdf
http://www.igi.tugraz.at/helmut/Presentations/AdaptiveControl
.html
http://www.irs.ctrl.titech.ac.jp/~dkura/ic2004/lec405.pdf
http://www.irs.ctrl.titech.ac.jp/~dkura/ic2004/lec406.pdf
http://mchlab.ee.nus.edu.sg/Experiment/Manuals/EE5140/ada
p1.pdf

Adaptive Control 2

Uploaded by

Copyright:

Available Formats

Adaptive Control 2

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Adaptive Control 2

Uploaded by

Copyright:

Available Formats

What are the two main ways that parameters can be adjusted in an MRAC?

What are the two main ways that parameters can be adjusted in an MRAC?

What are the main components of an MRAC?

What are the main components of an MRAC?

ADAPTIVE CONTROL

Adjustment of system parameters

Error criteria selected here is

GRADIENT or MIT rule.

The loss function can also be chosen as

first MRAC is implemented by this function

sign e sign (e)

For multivariable systems

APPLICATIONS FOR MIT

ADAPTATION OF A FEED FORWARD

MRAC FOR ADJUSTMENT OF FEEDFORWARD GAIN BY MIT RULE

The constant k for this plant is unknown. However, a

The cost function chosen here is,

The error is then restated in terms of the transfer functions

As can be seen, this expression for the error

Finally, the MIT rule is applied to give an expression for

To tune this system, the values of ko and gamma can be

NOTES ON DESIGN WITH MIT

OBSERVATIONS FROM GRAPH:

MRAC FOR A FIRST ORDER

1 & 2 are 2 adjustable parameters

B. DIAG FOR FEED FORWARD GAIN CONTROL USING LYAPUNOV RULE

The error equations are

The value of dv/dt is <=0 [which is ve semi

You might also like