| |
|
S18 Processos Mecânico-Metalúrgicos |
| |
|
Title:
|
NEURAL NETWORK REPRESENTATION OF ALEXANDER S MODEL FOR THE ROLLING PROCESS
|
| |
|
Summary :
|
|
ABSTRACT. THE ROLLING PROCESS MATHEMATICAL MODELING INVOLVES NONLINEAR PARAMETERS AND RELATIONSHIPS THAT USUALLY LEAD TO NONLINEAR EQUATIONS OF DIFFICULT NUMERICAL SOLUTION. SUCH IS THE CASE OF ALEXANDER S MODEL (1972), CONSIDERED ONE OF THE MOST COMPLETE REGARDING THE ROLLING THEORY. FOR SIMULATION PURPOSES, ALEXANDER S MODEL REQUIRES TOO MUCH COMPUTATIONAL TIME, WHICH PREVENTS ITS USE IN ON-LINE CONTROL AND SUPERVISION SYSTEMS.
IN THIS WORK, TWO NEURAL NETWORK STRUCTURES ARE TRAINED USING PROCESS AND OPERATION DATA RESPECTIVELY, GENERATED BY ALEXANDER S MODELS. THE NEURAL MODELS ARE VALIDATED THROUGH SIMULATION. FINALLY, THE NEURAL NETWORK MODELS ARE USED TO OBTAIN THE SENSITIVITY FACTORS OF THE PROCESS BY DIFFERENTIATING THE NETWORK OUTPUTS.
IT IS SHOWN THAT THE NEW NEURAL NETWORK REPRESENTATIONS ALLOW TO OBTAIN PROCESS EQUATIONS FOR DIFFERENT OPERATION POINTS. RESULTS OF THE REPRESENTATIONS ARE PRESENTED.
KEYWORDS: STEEL INDUSTRY, ROLLING PROCESS, NEURAL NETWORKS
|
| |
|
Author :
|
Gálvez, J.M. 0
Helman, H. 0
Zárate, L.E. 0
|
| |
|
Paper View :
|
|
COBEM99
| Organizing Committee | Keynotes
| Technical
Sessions | Author
| Symposia and special sessions
| Referees |
Work
Titles | Venue
|
