Information Technology Reference
In-Depth Information
the suggested approach and the adaptive neuro-fuzzy inference system applying the
control principle demonstrated that both strategies are appropriate with adequate
transient response short settling time, and minimal steady-state error as evaluated by
standard performance indices.
The transductive neuro-fuzzy inference system was also shown to be a simple,
fast, precise, and computationally viable tool for modeling and controling such com-
plex processes as drilling. Another advantage over other neuro-fuzzy systems is
that it makes acceptable predictions with few data points, permitting the dynamic
integration of new data without an explicit training stage.
7.5.2 Tuning of Fuzzy Controllers by Optimization
The main goal of this section is to apply the framework given by the cross entropy
method to the optimal tuning of fuzzy control systems. An optimal fuzzy controller
is designed and implemented for cutting force regulation in a network-based appli-
cation. The control algorithm is connected to the process via Ethernet. The output
(i.e. force) signal is measured with a dynamometer, and the control signal (i.e. feed
command) is transmitted through the network. In such a situation, a network induced
delay is unavoidable. The key issue is to derive optimal controller parameters that
yield a fast and accurate response with minimum overshoot through the minimization
of the integral time absolute error (
ITAE
) performance index. The resulting optimal
fuzzy controller should be able to deal with uncertainties and nonlinearities in the
drilling process, in addition to delays in the network-based application.
Once again it is necessary to point out the need of three key elements are:
•
A rough model of the plant, represented in this case study by a transfer function. In
this study the whole system is represented approximately as a third-order system,
and the transfer function can be represented as follows:
F
(
s
)
1958
G
P
(
s
)
=
=
(7.17)
f
(
s
)
s
3
+
17
.
89
·
s
2
+
103
.
3
·
s
+
190
.
8
where
s
is the Laplace operator,
f
is the command feed and
F
is the cutting force.
•
A performance index or performance indices, in this case study the ITAE (
7.13
)
performance index is applied. The main rationale for this choice has been already
discussed in Sect.
7.4.1
.
•
the initial fuzzy controller. The initial fuzzy controller is summarized as follows.
A typical fuzzy controller that describes the relationship between the change of
the control signal (e.g., the feed-rate increment
f
), and the error (e.g., force error
e
F
and its change
e
F
) is considered (Yager and Filev
1994
). This type of fuzzy
controller was suggested as early as the mid-seventies and remains a standard in the
field of fuzzy control (Mamdani and Assilian
1975
). The well-established steps for
defining input and output membership functions and for constructing fuzzy-control
Search WWH ::
Custom Search