Comparison of regression and artificial neural network models for prediction of delta ferrite content in stainless steel claddings

Article type: Research Article

Authors: Palani, P.K.^{a; *} | Murugan, N.^b

Affiliations: [a] Faculty of Mechanical Engineering, Government College of Technology, Coimbatore – 641 013, India | [b] Coimbatore Institute of Technology, Coimbatore – 641 014, India

Correspondence: [*] Corresponding author. Tel.: +91 422 2513080; E-mail: [email protected]

Abstract: This paper discusses modeling and prediction of delta ferrite formation during the cladding of 317L flux cored wire onto the structural steel plate using artificial neural network and regression analysis. Comparison between the two models is made. Data required for modeling were obtained from the experiments conducted using a central composite rotatable design of experiments. The study revealed that modeling of delta ferrite using neural network is roughly 2.5 times more accurate compared to modeling using regression analysis. Neural network and regression models are able to predict the delta ferrite content with an average percentage error of the order of 0.29% and −0.74%, respectively.

Keywords: Cladding, delta-ferrite, ferrite number, neural network, response surface methodology

DOI: 10.3233/KES-2006-10603

Journal: International Journal of Knowledge-based and Intelligent Engineering Systems, vol. 10, no. 6, pp. 433-443, 2006