Optimization-of-MIG-welding-using-Random-Forest-Regression

The idea of a sustainable approach has grown in popularity in the manufacturing sectors as manufacturers explore the potential for achieving economic, environmental, quality and social factors for improved performance of a manufacturing process. In this research, an effort has been made to analyse how the interaction of different welding parameters affect various responses using Regression analysis. Welding experiments were conducted with three multi-level dominant weld parameters employing the L20 Taguchi Orthogonal Array Design of experiments. An orthogonal array is a table of experimental conditions that is designed such that each column represents a different factor and each row represents a different combination of the levels of these parameters. Operating levels for parameters like gas flow rate, voltage and distance from weld were varied. Responses considered in the study include operator fatigue, time to weld, weight of spatter and slag, tensile strength, hardness and cost of power consumption. The analysis of the responses was done using Random Forest Regression technique. Random forest regression has the main benefit of reducing overfitting and offering more precise forecasts. The unique aspect of this work is the creation of a sustainability analysis of the MIG welding procedure used to weld mild steel