Neural Network Approach to Modeling the Laser MaterialRemoval Process

1
Neural Network Approach to Modeling the Laser
Material-Removal Process
By Basem. F. Yousef  London, Canada,
N6A 5B9
December 2001
2
Organization

• Introduction

• Experimental setup and data acquisition

• Neural networks concepts and models

• Model outputs and results

• Model validation

• Conclusions and recommendations

3
INTRODUCTION
4
Introduction
What is laser micro-machining ?
Laser micro-machining is the process of
manufacturing parts of dimensions from 0.1 ?m to
1000 ?m using the laser beam as a cutting tool.
laser-drilled orifices (all less than 100 µm in
diameter) in catheter tubing.
Why laser micro-machining?

• The global trend of industry is moving toward
  miniaturization
• Micro-scale parts are used in diverse fields such
  as medical bio-medical, microelectronics,
  opto-electronics, space and others.

Microgear of Al2O3 with 120 m m diameter,
produced by laser ablation (Courtesy of
Microlas).
5
Laser Micro-Machining System and Controlling
Parameters
6
Objectives

• To investigate and analyze how the geometry of
  the final surface profile forms and depends on
  the laser pulse energy.
• To develop an artificial neural network model,
  which can predict the laser pulse energy needed
  to produce a crater with specific depth and
  diameter on the surface of a specific material,
  and the expected variation in the produced crater
  depth and diameter associated with the modeled
  pulse energy.

7
Procedure
Utilizing a neural network involves

• Conducting experiments and acquiring data

• Developing the neural network models

• Training the networks using the experimental data

• Recreating outputs by the trained model

8
EXPERIMENTS
9
Experimental Setup and Data Acquisition
Crater parameters
The crater volume is calculated by
Sample picture provided by the surface profiler
V abhc
10
Variation of Depth for Craters Produced by Pulses
with Pulse Energy of 40.4 µJ
11
Crater Depth vs. Pulse Energy (Brass)
Mean
Crater depth - hc (?m)
Mean
Mean -
Pulse energy - E (?J)
12
Crater Average Diameter vs. Pulse Energy (Brass)
13
Mechanism of Material Removal by a Laser Pulse
14
NEURAL NETWORKS
15
Typical Multi-layer Neural Network
First hidden layer
Second hidden layer
Input signals
Output layer
Crater depth -hc
Laser Pulse Energy-E
Crater diameter -dc
Neurons
16
Basic Operation Performed by a Neuron
INPUT SIGNALS (xi)
BIAS
Neural Processing Element
y
X
hc
1
OUTPUT
Mapping
0
dc
X
y
Neural input space (vector)
Neural output space (scalar)
Nonlinear mapping function
Crater depth -hc
X
y
Ne
Laser Pulse Energy-E
Crater diameter -dc
17
Neural Network Model in Training Phase
Neural Network Modeler
Inputs
Modeled output
COMPARISON
CORRECTION
Actual output
In order to reduce the (error) difference between
the modeled output and the desired output, the
neural network updates its weight values by the
back-propagation algorithm. In this method, the
error signal originating at the output layer
neurons is back-propagated through the network in
the direction of the first layer and the weights
are updated to reduce that error.
18
Approximating a Continuous Function
1
1
Data points used for training

1
_
Approximate function
e

• A two-layer neural network can form an
  approximation to any continuous nonlinear mapping
• Training set consists of input-output pairs (x,d)

19
The Interconnection of the Artificial Neural
Networks for the Operation Mode.
Crater depth -hc
Laser Pulse Energy-E
Crater diameter -dc
ANN2
ANN1
20
MODEL OUTPUTS
21
Crater Depth and Diameter vs. Modeled and Actual
Energy (Brass)
22
Modeling the Variance of Depth and
Diameter (Brass)
Depth standard deviation vs. pulse energy.
Modeled
Actual
Diameter standard deviation vs. pulse energy.
Pulse energy (µJ)
23
Change in Diameter Under the Effect of Change in
Energy
Model outputs overlapping with experimental data
Model outputs superimposed on experimental data
points for verification and comparison
purpose. Nonlinearity is obvious when comparing
when E2-E with E-E1.
Mean depth-mean diameter curve
dc1 dc10
dc
Experimental data points
dc2 dc-10
Crater diameter dc (µm)
Model outputs falling outside experimental data
region are Modeled E for 80 dc. Modeled E for
50 dc.
Diameter increase
E1
E
E2
Pulse energy- E (µJ)
24
MODEL VALIDATION
25
3D Data Visualization
Elliptical regions confining the experimental
data areas associated with 3 energy levels.
26
Mesh Confining Experimental Data
27
Model Validation
110 mean diameter 105 mean diameter
Mean diameter 95 mean
diameter 90 mean
diameter 80 mean diameter
All simulation curves are inside the mesh except
80 mean-diameter curve. Curve A corresponds to
craters having depth 19.84 µm.
Curve A
Curve A intersects with simulation curve 80
mean diameter at the anticipated point of
intersection with a corresponding error of 2 .
28
Model Validation
Verification curves corresponding to same-depth
pulses are intersecting with model-output curve
80 mean diameter. (Numbers on the figure show
the depths of craters - hc, which belong to each
curve).
29
Multi-Material Model
Depth (hc)
ANN1
Diameter (dc)
Pulse energy (E)
Material property (k)
ANN2
30
Theoretical Equation for Volume of Material Melt
by a Laser Pulse
Tf Melting point.
T0 Ambient temperature.
Lf Latent heat of fusion.
Density.
R Surface reflectivity
CP Heat capacity
Material Property
Sensible Heat of Melting
31
Multi-Material Model Outputs
32
Multi-Material Model Outputs
copper
brass
Crater mean diameter dc (µm)
Modeled energy (brass) Actual energy (brass)
Modeled energy (stainless steel) Actual energy
(stainless steel) Modeled energy (copper)
Actual energy (copper)
Stainless steel
Pulse energy E (µJ)
33
Neural Network Approach to Modeling the Laser
Material-Removal Process

• Conclusions
• The developed neural network successfully modeld
  the actual process behavior to high degree of
  accuracy.
• The successful research results set the stage for
  valuable and promising future work in the field
  and for further improvement in process
  performance.
• Future Work
• Model the process outputs in terms of different
  input parameters such as focal spot, frequency
  and feed rate.
• Test the neural network capabilities to model the
  process when new materials (other than those used
  for training) are considered.

34
THANK YOU