Open Access Open Access  Restricted Access Subscription Access
Open Access Open Access Open Access  Restricted Access Restricted Access Subscription Access

On Dissimilar Welding of AISI 304 and EN 8 Steels through Metal Active Gas Welding : Part II-Estimation of Weld Characteristics Using Regression Analysis and Neural Networks


Affiliations
1 Department of Metallurgical and Materials Engineering Indian Institute Technology Kharagpur, Kharagpur, 700032, India
2 Department of Mechanical Engineering Kalyani Government Engineering College, Kalyani-741235, India
     

   Subscribe/Renew Journal


Nowadays, researchers have been using several predicting tools in the areas of defense, marketing, finance, and engineering. In the area of welding processes, estimation of response parameters is done. As a predicting tool in this investigation, artificial neural networks (ANN) and regression equations are used. Using the ANN model, predictions can be made through various learning methods possible with this algorithm. The regression equation for each response parameter is obtained from MINITAB software. Weld bead geometry, hardness, and maximum bending load of the welded zone are predicted. Sets of input and output data needed for experimental runs are obtained by joining AISI 304 and EN 8 steels together using the GMAW process. To predict weld bead geometry and mechanical properties of the weld zone of dissimilar steels, two separate prediction tools are used. The outcomes are then compared. Such research is novel in the field of predicting and comparing the output parameters of different weld joints using ANN and regression analysis (RA). It is concluded that ANN as well as regression equations have predicted the weld bead geometry, hardness, and maximum bending load with a little error. It is also found that ANN provides satisfactory predicted results with much less error than the results obtained from the regression equation.

Keywords

ANN, Regression Equation, GMAW, ANOVA, MATLAB, MINITAB.
User
Subscription Login to verify subscription
Notifications
Font Size


  • On Dissimilar Welding of AISI 304 and EN 8 Steels through Metal Active Gas Welding : Part II-Estimation of Weld Characteristics Using Regression Analysis and Neural Networks

Abstract Views: 354  |  PDF Views: 3

Authors

Tapas Bera
Department of Metallurgical and Materials Engineering Indian Institute Technology Kharagpur, Kharagpur, 700032, India
Santanu Das
Department of Mechanical Engineering Kalyani Government Engineering College, Kalyani-741235, India

Abstract


Nowadays, researchers have been using several predicting tools in the areas of defense, marketing, finance, and engineering. In the area of welding processes, estimation of response parameters is done. As a predicting tool in this investigation, artificial neural networks (ANN) and regression equations are used. Using the ANN model, predictions can be made through various learning methods possible with this algorithm. The regression equation for each response parameter is obtained from MINITAB software. Weld bead geometry, hardness, and maximum bending load of the welded zone are predicted. Sets of input and output data needed for experimental runs are obtained by joining AISI 304 and EN 8 steels together using the GMAW process. To predict weld bead geometry and mechanical properties of the weld zone of dissimilar steels, two separate prediction tools are used. The outcomes are then compared. Such research is novel in the field of predicting and comparing the output parameters of different weld joints using ANN and regression analysis (RA). It is concluded that ANN as well as regression equations have predicted the weld bead geometry, hardness, and maximum bending load with a little error. It is also found that ANN provides satisfactory predicted results with much less error than the results obtained from the regression equation.

Keywords


ANN, Regression Equation, GMAW, ANOVA, MATLAB, MINITAB.

References





DOI: https://doi.org/10.22486/iwj.v55i3.213078