Information Technology Reference
In-Depth Information
7.5 Case Studies
This section presents the design and implementation of learning and optimization
strategies for tuning fuzzy and neurofuzzy systems in the field of drilling process
control. The drilling process is selected is one of the most common machining oper-
ations used in manufacturing, comprising up to 50% of all machining work. Drilling
has had a major impact on production technology in many industries, such as the
automotive, die/mold and aerospace industries. However, it is one of the processes
that has received the least attention in regard to improvement through the applica-
tion of control techniques. Any improvement through new control and monitoring
systems is reflected in a higher production rate and increased economic efficiency.
Indeed, advancement in newcontrol systems can yieldmany benefits, such as reduced
cycle time, tool breakage prevention and cost efficiency, in addition to improved part
quality.
Metal-cutting operations, of which drilling is a subset, have been modeled in a
variety of ways. Many approaches explicitly model the cutting force through geomet-
ric and process variables, e.g. spindle speed, feed rate, depth of cut, etc. The main
bottleneck encountered in designing a control system for drilling is that process
dynamics cannot be accurately modeled mathematically.
Drilling force is the most important variable in the drilling process. Higher feed
rates increase the material removal rate along with an increase in drilling force.
However, higher cutting forces and drilling torques have negative effects as well,
such as rapid drill wear, tool vibration and the risk of catastrophic tool failure. Thus,
it is important not only to maintain a constant drilling force, but also to obtain good
closed-loop dynamic behavior without oscillations and overshoots, thus increasing
useful tool life.
The overall control architecture, including the machine tool that performs the
drilling process and the control system, is depicted in Fig. 7.7 . Drilling processes
are conducted in a machine tool equipped with an open computer numerical control
(CNC). A personal computer (PC1) is connected with the open CNC via a Profibus
network. PC1 performs three tasks. The first task is to measure force, F .Theforce
is directly measured from PC1 with a Kistler 9257B dynamometer at a sampling
frequency of 5kHz. Signal conditioning and filtering of the raw data are then per-
formed. The second task is communication with a second personal computer, PC2,
via Ethernet with standard CORBA middleware. Filtered force F and other parame-
ters and variables (e.g., spindle speed and depth of drill) are passed to PC2 in this
fashion. The third task is to receive the control action computed by PC2 with the data
interface and synchronization tasks performed by commercial software (Labview,
NC-DDE application) over Ethernet. PC2 cannot be connected directly to the CNC
due to proprietary CNC software constraints.
Search WWH ::




Custom Search