You are on page 1of 23

Repetitive Control Theory

Basics of Control Theory


Open Loop Versus Closed Loop
1 Open Loop systems are always stable unless plant behavior
is unstable where as closed loop system, if properly
designed can compensate for instability of plant for
example of balancing of Inverted Pendulum in closed loop.
2 Closed loop can improve transient response and
compensate for disturbances, nonlinearity of the plant etc.
3 Open loop control is simple where as closed loop is
complex as compared with open loop as controller design
requires additional hardware and has to be design for a
stable operation.

Goals for Design of a Controller in a Closed Loop


Control
Fast Transient Response
Zero steady state error
Should be stable over complete range of operation.

Example of Stability by Perturbation

5
V

15
V

1
0
K

1
0
K

5V

+5
V

+
5
V

+1
5V

5V

15V

10
K
10
K

+5
V

-5V

+5
V

+1
5V

Analog Controller Versus Digital Controllers


Implementation of analog controllers is quite old and well
known from time of steam engine in the recent history.
Analog Controllers have wide band-width
Easy Implementations.
Hardware size is proportional to number of functions to be
implemented.
Can be adopted for linear as well as non-linear plant
behavior.
Can not log/save Data
Fault conditions will be lost on power failure.

Advantages of Digital Controller


Some of the advantages of digital controller over
analog controller are listed below.

Flexibility : Any changes in methodology or control


parameters dont require changes in hardware.

Advanced Control : Due to implementation of digital


control, It has been possible to implement more
advanced control schemes such as Dead-Beat control,
Disturbance observer, Repetitive control, Kalman Filter,
Park and Clark transforms etc. to achieve better
performance. Can be designed for linear as well as nonlinear plant behavior.

Communication & Control : System can


communicate with several other systems as well as
with master controller to implement factory
automation. This helps in setting control parameters
remotely.

Reduction in Hardware size of controller :


As one DSP chip can be used for generating
several PID loops and signal monitoring & control
hence require minimal system

Data Logging and Fault recording: Inevent of fault


condition all the data and fault conditions can be
recovered for further action
Multiplexing and Multitasking : With general
purpose digital hardware, many more signals can be
sampled by just adding a multiplexers which increases
the flexibility of the control and one DSP chip can
perform several operations such as computation of
control law, signal sampling, fault recording,
communication etc. Fault recording could be
summarized as one of the best facility that digital
control can offer.

Disadvantages of Digital Controller


Sampling Effect : Unless sampled at higher rates can
result into aliasing of signal and there by loosing
important data

Time Delay Effect : Sampling introduces delays. Delay


introduces frequency dependent phase lag with no
change in amplitude and Phase lag reduces closed-loop
stability

Quantization and Delay Effects : A-to-D conversion


causes quantization errors. This reduces accuracy of
measurement of input/output signals. Effect of this can
be reduced by higher order A/D converters.
Effect of Time Delay on Step Response and Stability

Some Popular Techniques Used In Digital Control

Dead-Beat Control
Predictive Control
Digital PID
Observer based Sensor-less Control
Repetitive Control Etc

Closed-loop system

If we assume that GH >> 1, then the overall transfer


function simplifies to

error

1 1 0
if GH>>1
1 GH GH

If reference input is a DC quantity then using PI loop


will give zero steady state error.

+-

Kp

K
I
S

Gai
n
10
Hz

100H
z

But for AC Reference Input, PI loop gain has finite


gain and can not track the AC input hence to make
the controller track exactly to the given AC reference,
resonant controller scheme was introduced.

but then with with the advent of microprocessors dq


transformation technique was developed. This uses
conventional PI loop and AC quantities are
transformed in to DC to get zero steady state error.

Principal of Repetitive Control

Repetitive control intends to track/reject


arbitrary periodic signals/disturbance of a
fixed period

Tracking/Disturbance rejection of periodic


signals appear in many applications

Hard disk/CD drives


Electric power supply
Robotic motions

Steppers in IC productions
And many others

History: The First Example


Magnet power supply for a proton
synchrotron (Nakano and others) for
Ring Magnet

Control Objective:
Control the power supply curve
(periodically) to the following shape:
Precision requirement: order of 0.1V!

Internal Model Principle of a Control System


Vin(s
)

GC(S
)

According to internal model principal if Laplace transform of


Vin(s) and Gc(s) should have same poles to exactly track the
reference.
If input signal or the disturbance is periodic then it can be
represented by Fourier series as
So the minimal system Gc
will consist of following poles.

(n ) 2

0
Gc (s) 1
..................(1)
s n1(s 2 n 2 2 )
0

sin(s' )
(s' ) 2 (s' ) 4 (s' )6
1

...........(2)
(s' )
3!
5!
7!

This polynomial has roots at s = +/- 1, +/- 2, and so on.


So
sin(s' )
s'2
s'2
s'2
(1
)(1
)(1
)..........(3)
2
2
2
(s' )
1
2
3

Expanding right-hand side we get


2 2 2
sin(s' )
[1 ( 1 1 ...)s'2 ( 1 1 1 1 ...)s'4 ...... s' s' s' ...]...(4)
s'
12 22
12 2 2 22 32
12 22 32

Comparing coefficients of equn 2 & 4 we get following.

1 1 1
2
1 ...........
2 2 2
6
2 3 4

and similarly

1 1 1 1 1 . 1 ........ 2
120
22 32 32 42 42 52
In any given function we may replace a variable with
another variable with out changing its value.
Let be s = j*s in equn 4 then we get

sin( js)
s2
s2
s2
(1
)(1
)(1
)..........(5)
2
2
2
( js)
1
2
3

sin( js) (1 s 2 )........(6)

2
js
n 1 n
sinh s (1 s 2 )........(7)

2
s
n 1 n
s

sinh s n 1

1
.........(8)
2
s
(1 2 )
n

sinh s s n 1

0T

1
.........(9)
2
s
(1 2 )
n

2 1
sinh s s n 1

1
.........(10)
2
s
(1 2 )
n

2 1
sinh s s n 1
0

1
.........(11)
2
s
(1 2 )
n

Again substitute S -> S/0 then


Then
T

1
1
2

. Gc (s)........(12)

s s
2
sinh( ) n 1 (1 s )
0
n 20 2
But
So
T

2
Gc (s)........(13)
sT
sinh( )
2

To calculate the transfer function of the above equn first we


have to substitute

s = j and T = 2/ 0 and then we

get

Te

(1 e

j2 and
0

Te
(1 e

j 2

T
sin(

)
0

e
1,0

j 2
0

(1 e

j 2
0

So the minimal system for Gc(s) is given as below.

This can be derived by using following block.


Verror(
s)

=
Gc(s)

or
Verror(s
)

Verror(s)

Periodic Waveform
Generator

=
G
c(
G
s)
c
Periodic
Waveform
(
Pl
Generator
s
a
)
nt

P
l
a
n
t

P
l
a
n
t

G
c
(
s
)

2
T

3
T

=
G
c(
s)

=
G
c(
s)

Pl
a
nt

Pl
a
nt

error

++

z-N

ZK

Q(z-1)

Repetitive Control
Block

+10V

Repetitiv
e
Control

erro
r + +

e-sT
Repetitive Control
Block

+10V +

e-sT

+10V +

+5V

+5V

5V

e-sT

0V

+10V +

+2.5
V
+

7.5V

+7.5
V

+
+5V

+10V +

+1.2
5V +

8.75
V

e-sT

+8.7
5V

+
+7.5
V

e-sT

You might also like