الفهرس 
صفحة العنوانOnly 14 pages are availabe for public view
 |  | from | 133 |  |  |
| | | from | 133 | | |
Abstract
br>
Metal deposition is one of the promising fields. Until now metal deposition is
used for shape modelling. Materials and sizes are so limited. To get rid of
these limitations welding robots were used instead of laser 3D printers, and
instead of metal powders MIG welding was used. But to make this process a
full production and prototyping one optimizing mechanical properties of
products was required.
To enhance accuracy of products width of deposited lines needed to be
minimal. Also key to make mechanical properties acceptable hardness
required to be close as possible to base metal one. So these were optimization
targets.
As MIG is one of the highly nonlinear processes NN were used to model it. 3
optimization techniques were used before developing 4th new one by
optimizing ACO initial parameters values which is one of the main problems
facing ACO optimization.
First stage was to develop models for the MIG process. Current, voltage and
travel speed are inputs and deposited line shape (width, penetration and
height) and hardness of 3 levels on each line are outputs. Values of inputs
were gotten from wires catalogues and suppliers applications as ranges.
Second stage was optimizing the process to get optimal parameters that
satisfies optimization targets. Using the experience from optimization by HS
and ACO the fourth one was made and applied on the process. Results from
the 4 optimization practically tested. The newly developed one showed
performance very close to the HS which was the best one among the previous
3. Also ACO, HS and modified ACO testing results were close to those got
from computer in contrast of those for genetic algorithm.