文档介绍：International Journal putational Cognition (.htm)
Volume 2, Number 1, Pages 79–112, March 2004
Publisher Item Identifier S 1542-5908(04)10104-8/$
Article electronically published on February 1, 2003 at . Please cite
this paper as: hMoamar Sayed Mouchaweh, “Diagnosis in Real Time for Evolutionary Processes in
Using Pattern Recognition and Possibility Theory(Invited Paper)”, International Journal pu-
tational Cognition (.htm), Volume 2, Number 1, Pages 79–112, March
2004i.
DIAGNOSIS IN REAL TIME FOR EVOLUTIONARY
PROCESSES IN USING PATTERN RECOGNITION AND
POSSIBILITY THEORY(INVITED PAPER)
MOAMAR SAYED MOUCHAWEH
Abstract. In this paper, we propose to use the evidence classifica-
tion method Fuzzy Pattern Matching (FPM) to realize the diagnosis
in real time. Then we show how the integration of the incremen-
tal learning in FPM allows plete the missing knowledge in
the database and to follow, in real time, the changes on the shape
of classes. Finally we develop the incremental learning approach to
predict the evolution of a process from normal to abnormal operating
state. The goal of this prediction is to protect the system from the bad
consequences of abnormal functioning and thus to realize a conditional
maintenance instead of a systematical one. Copyright °c 2003 Yang’s
Scientific Research Institute, LLC. All rights reserved.
1. Introduction
Monitoring an industrial process, at each instant, increases the produc-
tivity and decreases the production costs. When a fault is detected, the
monitoring system executes a correction procedure to bring back the process
in normal operating conditions. A monitoring system involves the following
steps : (1) data acquisition and preprocessing, (2) diagnosis and (3) control.
The data acquisition is realized in using a set of sensors. These sensors
provide signals which are analyzed to extract the useful characteristics in
order to represent the process behavior into tow operat