PREDICTION AND OPTIMIZATION OF MIG WELDING PROCESS PARAMETERS FOR BEAD GEOMETRY USING ANN AND GENETIC ALGORITHM

Abstract

Welding is the most practical and productive approach to join metals for all time. It is the main method for joining at least two bits of metal to make them go about as a solitary piece. Almost everything we use today which we call them as necessities such as car which is our mode of transport, mobile phones, metal doors and all other stuff made of metal involve some sort of welding. It has been classified in number of processes and one of them is Metal Inert Gas welding(MIG) also known as Gas Metal Arc welding(GMAW) by American welding society. This report examines the research work done in the prediction and optimization of process parameters of MIG welding for bead geometry using ANN and GA. MIG welding also known as MAG welding by American Welding Society is a form of arc welding in which a solid electrode wire is used as an electrode and is feed by a spool mounted on the welding gun or torch. Shielding gas is released through the nozzle in the nozzle from an eternal high-pressurised cylinder. The gases used for the shielding purposes are mostly inert in nature. The material selected to carry out the study is AA6082. We all are aware of the fact that various welding parameters govern the welding process and each of them affects the property of the welded joint in a different way. The process parameters selected for the study are welding voltage, gas flow rate and wire feed rate. The response parameter is bead geometry i.e. bead width, reinforcement height and penetration. Design expert software is used to create the design matrix using central composite design for doing the experiment. After the experimentation, samples are created by polishing the cross sectional area using emery paper. Then bead geometry is measured using Image J software. After measurement, prediction of bead geometry is done using the Matlab software and percentage error is calculated. For obtaining optimum value of bad geometry, optimization is done by genetic algorithm using Matlab software. Optimization of process parameters is also done using vi RSM and then compared with the result of genetic algorithm. For result analysis, perturbation plots and 3D surface response graphs are plotted to study the interaction effect of various process parameters on the response parameters.