Home // INTELLI 2015, The Fourth International Conference on Intelligent Systems and Applications // View article

Process State Observation Using Artificial Neural Networks and Symbolic Regression

Authors:
Susanne Fischer

Keywords: Observation; control; reduction; regression; manufacturing

Abstract:
Process state observation is important for efficient automated online control in manufacturing. In this paper, we propose a new concept for online observation of the process state. First, we obtain state variables of physically-based numerical simulations of a process. After that, we reduce the state variables to a few characteristic features using artificial neural networks. As a result, we have a process state representation in a lower dimensional feature space. Using this representation, we apply symbolic regression to find a mathematical model that describes the state in the feature space. By applying these methods to a cup deep drawing process, we can describe 99% of the state variables' variation using only 7 features instead of 400. For these 7 features, we can find mathematical descriptions that represent a reduced process model which is used for process state observation.

Pages: 142 to 147

Copyright: Copyright (c) IARIA, 2015

Publication date: October 11, 2015

Published in: conference

ISSN: 2308-4065

ISBN: 978-1-61208-437-4

Location: St. Julians, Malta

Dates: from October 11, 2015 to October 16, 2015