Illustrations
To demonstrate neural network approach, a training set of example process plans was generated for spur gear with five features: a hole, a keyway, and two faces or identical in tolerances and dimensions, and the gear teeth. Each feature was associated with a set of attributes, defined in given table. For each illustration, the value for the attributes was randomly assigned inside common manufacturing ranges for this feature, subject to physical constraints.
Table of: Example Features and Associated Attributes
|
Feature
|
Attributes
|
|
Hole
|
Depth, diameter, size tolerance, position tolerance, circularity, straightness, surface finish
|
|
Face
|
Diameter, thickness, size tolerance, parallelism, surface finish
|
|
Keyway
|
Width, depth, inset, size tolerance, position tolerance, surface finish
|
|
Gear
|
Diameter pitch, error in action, surface finish
|
generated artificially by using rules. These rules after that constituted the domain transformation functions to be learned by the network. Above diagram shows some rules utilization to generate the operation sequences for hole features. Notice that these rules are non-trivial and, actually, need evaluation of relationships in between attributes. Following Figure shows some demonstrates process plans generated.
Table defines the machining operations considered to a network. A number of these operations, as like milling and honing, are utilized in the manufacture of more than one feature types.
If (depth/diameter ratio) >=3 then
If (diameter > 2) then "center drill" "trepan"
else
else
end if
"gundrill"
if (((diameter, 0.75) and
((size tolerance <=0.003) or
(position tolerance <=0.005))))
else end if
then
"central drill" "twist drill"
"twist drill"
if (straight <=0.001)
then
end if
"ream"
if (circularity <=0.001)
then
end if
"counter - bore"
if (surface finish <=16)
then
end if
"hone"