Transcript Model Based Control

•The initial feed-forward set point is obtained
from the optical power modeling done in
MATLAB.
Manual
Fiber-Fiber Alignment
•This set point is given as an input to the PMDI
motion control software which follows a PID loop
by measuring the power.
Semi-Automatic
 No standard for OE packaging and assembly
automation.
•We start with an initial guessed value of “a”
which is the center-to-center separation between
the slits.
Packaging is critical to success or failure of optical
microsystems.
 60-80 % cost of optical component/system is in
packaging.
For the double slit aperture, the irradiance at any point in space is
given as:
Automation is the key to high volume, low cost, and high
consistency manufacturing ensuring performance, reliability,
and quality.
kb
1 x
2 ka
1 x
I ( x)  A sin c ( sin (tan ( ) ) ) cos ( sin(tan ( )))
2
z
2
z
u (t )
Input
•The inner PID loop is repeated 5 times after
which the outer learning loop comes into effect.
•The learning loop updates the estimated set
points and tracks the actual set point. In this
simulation, the learning algorithm was run 28
times.
2
 = wavelength = 630 nm
k = wave number associated with the wavelength
a = center-to-center separation = 32 um
b = width of the slit = 18 um
z = distance of propagation =1000 um
y (t )
Real System
Output
+
Adjustment
Scheme
e(t )
yˆ (t )
Error -
Estimated
model
Model
We present the above optical system with one unknown( slit width -“a”)
exhibiting input-output differential equation y  my  Ku ( m is unknown and K
is known )
The variables u, y, and y are to be measured
{ y  x1 and y  x 2 }
Step 2: xˆ  0 1  xˆ   0  uˆ

ˆ  
{ Assume estimated model and e  x  xˆ }
0  m
Inaccurate modeling could lead to deviation from the actual
values.
 Activated at a lower sampling frequency.
 Specific and appropriate tasks.
 Provides opportunities for the system to improve upon its
power model.
 Adjust the accuracy on the basis of “experienced evidence.”
Optical setup
Step 1: x  0 1  x   0  u

  
0

m

 K 
K 
e1
Step 3: The Sensitivity coefficients S 
mˆ
are contained in
Amplifiers, Encoders
Interpolators, Motion Controller
ˆ
e
x
where e   y  yˆ y  yˆ  ,



mˆ
mˆ
T
Power Meter readings
̂
Learning Equation : m
 eS T e
We track “m” to be 1.86 which relates to a
slit width “a” of 32um.
Visit www.ece.drexel.edu/opticslab/results/Automation.wmv to watch the Model Based Control Video
Visit www.ece.drexel.edu/opticslab/results/learning.wmv to watch the Learning Identification Video
Criteria for choosing e
If e is too large, the schemes will diverge.
Initial X Encoder Position = 10477
Final X Encoder Position = 10810
If e is too small, then m̂will approach m
very slowly.
Initial Y Encoder position = 24
Final Y Encoder position = 25
Selection of a suitable e determined by a
trial and error process.
Comparison of Power Levels
Step 1: Assume system to be described as y  f ( y, u,  ) , where y
is the output, u is the input and is the vector of all unknown
parameters.
1) Current State-of-the-Art = 0.644 uW
2) Our Technique: Model Based Control = 1.55 uW
Step 2: A mathematical model with the same form, with different
parameter values ˆ is used as a learning model such that yˆ  f ( yˆ , u, ˆ )
3) Our Technique: Learning Identification Control = 1.675 uW
Step 3: The output error vector, e , is defined as e  y  yˆ .
Thus, we show an increase in power
level reached along with increased
efficiency and accuracy.
Step 4: Manipulate ˆ such that the output is equal to zero.
 e  
 ˆ
ˆ
Step 5:It follows that e  e(  ) and e  
ˆ




1.
2.
3.
4.
Shubham K. Bhat, T.P.Kurzweg, Allon Guez "Learning Identification of Opto-Electronic Automation Systems", IEEE Journal of Special Topics in Quantum
Electronics, May/June 2006.
Shubham K. Bhat, T.P.Kurzweg, Allon Guez,” Simulation and Experimental Verification of Model Based Opto-Electronic Packaging Automation”, International
Conference on Optics and Opto-electronics Conference, Dehradun, India, December 12-15, 2005.
Shubham K. Bhat, T.P.Kurzweg, Allon Guez, “Advanced Packaging Automation for Opto-Electronic Systems”, IEEE Lightwave Conference, New York, October 2004
.
T.P. Kurzweg, A. Guez, S.K. Bhat, "Model Based Opto-Electronic Packaging Automation," IEEE Journal of Special Topics in Quantum Electronics, Vol.10, No. 3,
May/June 2004, pp.445-454.