Modelling on CVN toughness of weld deposits

The Charpy V Notch toughness (CVN) of steel is an important property while considering structural and heavy loading conditions. In welded structures, CVN is attributed to many variables like composition of steel, heat input of welding, pre- and post-heat treatments of the weldment, type of weldi...

Full description

Saved in:
Bibliographic Details
Main Authors: Chauhan, Rudrang, Nanavati, Purvesh, Pandit, Vinaykumar, Sharma, Shashank
Format: Article
Language:English
Published: Penerbit Universiti Kebangsaan Malaysia 2022
Online Access:http://journalarticle.ukm.my/20333/1/13.pdf
http://journalarticle.ukm.my/20333/
https://www.ukm.my/jkukm/volume-3404-2022/
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The Charpy V Notch toughness (CVN) of steel is an important property while considering structural and heavy loading conditions. In welded structures, CVN is attributed to many variables like composition of steel, heat input of welding, pre- and post-heat treatments of the weldment, type of welding process etc. The regression analysis works accurately for three to four variables. The property of weldment is associated to more than three-four variables. So this conventional regression analysis couldn’t capture associated trends among the variables due to their non-linearity. This complexity is countered well by artificial neural network (ANN) modelling. In the present work, artificial neural network approach is utilized for the prediction of CVN of ferritic steel welds, which is multi-phase complex engineering material. The multilayer perceptron (MLP) method is used for formulating the neural network models. Numerous models were made by adjusting the hyperparameters and a best model was selected having least training error. The crucial results obtained from this model where analysed from response graphs and contour plot. This (MLP) approach for formulating neural network model was proved to be efficient after validation procedure and the same model could be exploited well for predicting accurate value of CVN in a very time and cost-effective manner.