Int. J. of Mechatronics and Manufacturing Systems   »   2017 Vol.10, No.4

 

 

Title: Prediction of formability of adhesive bonded sheets through neural network

 

Authors: V. Satheeshkumar; R. Ganesh Narayanan; Deepak Sharma

 

Addresses:
Department of Mechanical Engineering, Amrita School of Engineering, Coimbatore, Amrita Vishwa Vidyapeetham, Amrita University, India
Department of Mechanical Engineering, IIT Guwahati, 781039 Guwahati, India
Department of Mechanical Engineering, IIT Guwahati, 781039 Guwahati, India

 

Abstract: The present work aims to predict the formability of adhesive bonded sheets accurately. The difficulty during incorporation of adhesive and adhesion properties accurately in finite element (FE) simulations while predicting the formability is addressed. Here an artificial neural network (ANN) model is developed based on the experimental data of adhesive bonded sheets which inherently includes actual properties of adhesive and adhesion. Feedforward back propagation algorithm is used for predicting forming limit from tensile test and cup height from deep drawing process. In FE simulations, thickness heterogeneities with factor 'f' have been designed in the base materials to predict the forming limit without adhesive and adhesion properties. The ANN results are validated through experimental results and also compared with FE results. A good correlation between experimental and ANN predicted results, and a considerable variation with FE results confirm the viability of ANN for predicting the formability of adhesive bonded sheets accurately.

 

Keywords: manufacturing; adhesive bonded sheets; neural network; numerical prediction; tensile behaviour; deep drawing; forming limit; cup height; feedforward back propagation algorithm; thickness heterogeneity; adhesive properties.

 

DOI: 10.1504/IJMMS.2017.10009919

 

Int. J. of Mechatronics and Manufacturing Systems, 2017 Vol.10, No.4, pp.321 - 354

 

Submission date: 05 May 2017
Date of acceptance: 11 Sep 2017
Available online: 22 Dec 2017

 

 

Editors Full text accessAccess for SubscribersPurchase this articleComment on this article