0
Research Papers

Identification of Die Thermal Dynamics Using Neural Networks

[+] Author and Article Information
Jaho Seo, Amir Khajepour

Jan P. Huissoon

Department of Mechanical and Mechatronics Engineering,  University of Waterloo, 200 University Avenue West, Waterloo, ON, N2L 3G1, Canadajph@mecheng1.uwaterloo.ca

J. Dyn. Sys., Meas., Control 133(6), 061008 (Nov 11, 2011) (9 pages) doi:10.1115/1.4004045 History: Received November 02, 2009; Revised March 16, 2011; Published November 11, 2011; Online November 11, 2011

The objective of this research is to identify a dynamic model that describes the temperature distribution in a die with uncertain dynamics using a neural network (NN) approach. By using data sets obtained from a finite element analysis (FEA) of the thermal dynamics of a die and applying NN off-line and on-line learning algorithms, the die model is identified. This identification approach has been conducted assuming fully measurable and partially measurable states. For the latter, a NN based adaptive observer is employed to estimate unmeasurable states. It is shown that the complex behavior of the die system with cooling channels can be accurately identified in both cases of fully and partially measurable states.

FIGURES IN THIS ARTICLE
<>
Copyright © 2011 by American Society of Mechanical Engineers
Your Session has timed out. Please sign back in to continue.

References

Figures

Grahic Jump Location
Figure 1

Shot block (a) and casting die (b)

Grahic Jump Location
Figure 2

Conformal cooling channel of shot block

Grahic Jump Location
Figure 3

Procedure for system identification using NNs

Grahic Jump Location
Figure 4

Monitored points for FEA

Grahic Jump Location
Figure 5

Operating input conditions for FEA

Grahic Jump Location
Figure 6

Temperature changes at node 1,2,3, and 4

Grahic Jump Location
Figure 7

NN structure for off-line training with LM algorithm

Grahic Jump Location
Figure 8

NN based adaptive observer model

Grahic Jump Location
Figure 9

Off-line training results using LM algorithm

Grahic Jump Location
Figure 10

On-line training results using on-line learning algorithm with optimized learning rate

Grahic Jump Location
Figure 11

Verification of performance of NN with adaptive observer using testing data set.

Tables

Errata

Discussions

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related eBook Content
Topic Collections

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In