27-06-2012, 12:31 PM
Automatic Train Operating System based on Predictive Fuzzy Control
Automatic Train Operating System.PDF (Size: 481.64 KB / Downloads: 52)
Introduction
In recent years, atteqts are being made to
realize autamtic operation utilizing microccmputers
in lieu of human cperators in such fields as
industrial plant and transportation system. Though
the canpter control is capable of Wing accurate
and quick response, its quality is s c " s inferior
to that of the skilled operators.
control was proposed by Prof. Mamdani2 as a mthcd of
algorithmication of the skilled operators' know-how
employing fuzzy set theory, arid is applied to m t
kiln and water filtration plant, etc.
mthd, however, realization of a control system,
which takes desirable states of the system being controlled
into consideration, had been difficult.
Predictive Fuzzy
Skilled operators have high-grade exprience
acquired through nurrrerous number of times of
-rating a system and perform high-quality control
satisfying the purpose of the system.
control ccmnand U based on objectives evaluation of
the control results is the predictive fuzzy control
system.
control is described below.
To define a
The algorithm for the predictive fuzzy
In the predictive fuzzy control, the control can-
"I U is made to be discrete value (u=Cl,C2 e),
and x and y are =de to be performnce indices of
control. Evaluations of x and y (e.g. gccd, bad)
are defined by the fuzzy sets distinguished by mrtlbership
functions +(XI and pi(y).
periodically assesses a control rules expressed by
the following sentence ; "If the performance indices
x and y are Ai and Bi respectively when the control
cormand U is decided to be Ci, then this control rule
is selected and the control comnand Ci is decided for
output of the controller."
trol rule is fonnrlated as follows.
Control Rules of Fuzzy AT0 Controller
The mrnbership functions of fuzzy sets
corresponding to the established perfomce idices
and the control rules are sham in Fig.4. The functions
and control rules are established by our
Wrience of developing automatic train operation
systems thus far.
that can be selected by the AT0 are limited discrete
value, i.e. Seven steps each for powering and
braking plus cmsting and mgency braking steps.
The following control rules are laid dam which stipulate
the details of evaluation for candidate
control c d s re garding the above mentioned two
control mes, i.e. CSC and TAX.
[b]Realization of Fuzzy AM Controll
A control technique for the predictive fuzzy
control was incorporated into an on-board-the-car AT0
equipment to serve as a program for microcomputer.
On this occasion, fundamental hardware of the equip
ment ren-ained unchanged, and only software was
dified to a c d t e to the fuzzy control.