Radial Basis Neural Network Based CAD System for Welding Material

Abstract:

The formula of industrial welding material is directly related to their physical properties, in order to study the relationship between the two factors. A computer-aided design system for welding materials based on RBF (Radial Basis Function) neural network theory is designed and implemented. The system is able to establish the artificial neural network model for impact factors and performance index, based on which the performance prediction, formula prediction and analysis modules can be generated. The simulated results show that the predicted results enjoys related error less than 5% compared with the measured data, which have great practical significance to guide welding material researchers to explore the correlation between impact factors and performance index.

• Periodical: Advanced Materials Research (Volume 710)
• Pages: 727-730
• DOI: https://doi.org/10.4028/www.scientific.net/AMR.710.727
• Online since: June 2013
• Authors: Juan Wang, Si Yu Lai
• Keywords: Computer Aided Design, Feed Forward, Feedback, Radial Basis Neural Network, Welding Material

References

[1] Gu X.S., Yao R.G., Kong H.Y. and Liu Q. Application of Artificial Neural Network in Welding Technology [J]. Development and Application of Materials, 2010, 35 (4): 83-87.

[2] Wassermam P.D. Advanced Methods in Neural Computing [M]. New York: Van Nistrand Reinhold Press, 1993.

[3] Fan S.H. and Li D.W. The Application of Neural Network in Forecasting the Efficiency of Utilizing Energy [J]. Journal of Guangxi University of Technology, 2006, 41 (4): 79-84.

[4] Ma J. Discussion on Application of CAD Technology in Mechanical Design [J]. Coal Technology, 2011, 38 (5): 23-27.