Experimental Investigation for Energy-Conscious Welding Based on Artificial Neural Network

The selection of input parameters in joining processes has remained a crucial task due to the energy-intensive behavior of welding processes. Low carbon alloy steel is the most widely welded material in the industry. The Manual Metallic Arc Welding (MMAW) of mild steel is most well-known among all welding procedures, as it offers a low-cost remedy, finds extensive use in structural work, restoration, & maintenance. The current study focuses on selecting suitable MMAW parameters for welding mild steel, taking into consideration power and joint quality as the decisive factors. The experiments conducted were designed using Minitab 18 software. The transverse tensile strength, joint hardness, and the impact energy of the joint region are measured as quality parameters of the welded joint. Measurement of power has been conducted using a power analyzer. Artificial Neural Network is utilized for training the network using data acquired from the experiments carried out. The regression model was applied in Matlab R2019a to determine the relationship between the input and result variables to greatly help predict the perfect mix of joint input parameters.